Geographic variations in opioid overdose patterns in Pennsylvania during the COVID-19 pandemic

Abstract

The convergence of the opioid epidemic and the COVID-19 pandemic has created new health challenges throughout the United States. Since the onset of the pandemic, media attention and scholarly research have drawn attention to the intersections of addiction and COVID-19. However, there remain few empirical studies that examine the direct impacts of the COVID-19 pandemic for opioid overdose patterns. Even fewer have integrated quantitative and qualitative methods to detail the place-specific dynamics shaping opioid overdose and addiction treatment during the COVID-19 pandemic. This article measures and maps change in the age-adjusted rate of opioid-related overdose incidents at the county level from 2018 to 2020. These analyses are combined with interviews conducted since December 2020 with public health providers in the state of Pennsylvania to identify the key factors influencing opioid misuse and transformations in addiction treatment practices.

1. Introduction

The U.S. opioid epidemic represents one of the most pressing societal challenges in recent decades. According to the Centers for Disease Control and Prevention (CDC), from 1999 to 2019, almost 500,000 people died from an opioid overdose (CDC 2021a). While national and state mitigation efforts have shown some success in reducing the distribution of opioid related mortality, as reflected in the distinct geographical patterning of opioid related mortality from east to west (Kiang et al., 2019), progress in fighting the opioid crisis has been challenged by the recent emergence and spread of the COVID-19 pandemic. Preliminary data released in 2021 by the CDC revealed a surge in opioid-related overdose mortality during the pandemic, with the biggest spikes observed in April and May of 2020 (Ahmad et al., 2021), coinciding with widespread state level stay-at-home orders, school closures, and the major global economic downturn triggered by the pandemic. This surge was driven primarily by illicitly manufactured fentanyl and other synthetic opioids (Ahmad et al., 2021), though recent evidence based on clinical drug testing data suggests that more dangerous drug combinations, specifically fentanyl combinations, are also contributing to opioid use patterns (Niles et al., 2021).

As varying state level responses to COVID-19 extended across the U.S. (e.g., shelter in place orders, social distancing guidelines, changing take home medication for opioid use disorder (MOUD) guidelines), growing unemployment rates and broader economic impacts associated with the pandemic created an environment that made those with opioid use disorder (OUD) generally more susceptible to disease, while compounding vulnerabilities for socially marginalized populations and underserved urban-rural counties (Glober et al., 2020; Rodda et al., 2020; Slavova et al., 2020; Rosenbaum et al., 2021). While a growing body of literature suggests important geographical trends in opioid overdose prior to the pandemic, with higher rates in rural than urban counties from 2007 to 2015, and higher in urban than in rural counties from 2016 through 2019 (Hedegaard and Spencer 2021), few studies to date have evaluated how geographic variance in opioid overdose rates and local response measures are potentially changing during the COVID-19 pandemic.

Additionally, the role of place-specific dynamics in shaping particular opioid overdose patterns are uncommon in the scientific literature, even though attention to these patterns would offer critical insights for public health research and policy. Some of the key factors contributing to OUD, such as economic stress, social isolation, and access to addiction treatment services, vary geographically and are produced by social, cultural, and political-economic processes. Attention to these factors and processes across space is urgently needed, especially in geographic areas that were experiencing high rates of substance use and misuse prior to the pandemic, such as the state of Pennsylvania.

To begin to fill this gap, this article examines change in the rate of opioid-related overdose at the county level in Pennsylvania coupled with the experiences of OUD health providers from 2018 to 2020. The pandemic and associated shutdown measures have challenged economic stability and opioid mitigation strategies in the state. Reflecting this, between March and April 2020, unemployment rates in Pennsylvania increased from 6% to approximately 16% (US Bureau of Labor Statistics 2020). Due to the ongoing economic recession, projected state revenue for the 2020 fiscal year was expected to fall by 9%, with even higher revenue losses projected for 2021 (National Conference of State Legislatures 2020). Amid the colliding health and socioeconomic crises, this article specifically examines how the COVID-19 pandemic is intersecting with county-level patterns of opioid overdose, in addition to public health responses in order to improve policy and mitigation efforts across the state.

The next section of the article provides an overview of the opioid epidemic in the United States and the ways it is being impacted by the COVID-19 pandemic. This is followed by a discussion of the case study and methodology, which integrates spatial analyses of opioid overdose patterns at the county level from 2018 to 2020 with interviews with public health providers in the state of Pennsylvania. We report the main research findings and then examine how these contribute to geographic research on infectious disease and addiction patterns. This is intended to inform policy responses designed to mitigate the spread of COVID-19 while simultaneously providing addiction treatment for populations misusing opioids.

2. COVID-19 and opioid misuse in the United States

The COVID-19 pandemic has exposed existing social determinants of health (Dasgupta, Beletsky and Ciccarone 2018) that likely generate geographic differences in opioid use and misuse. The history of the opioid epidemic in the U.S. demonstrates that its main patterns have been experienced unevenly across different social groups and geographies (Bernard et al., 2018; Santoro and Santoro 2018; Lyden and Binswanger 2019). Since the early 1990's, the opioid epidemic has gone through several waves, transitioning from a situation in which prescription opioids drove high death rates, particularly for white rural populations, to an epidemic characterized by heroin utilization largely among urban racialized populations, to one now characterized by synthetic opioids, particularly illicitly manufactured fentanyl that is utilized across a broad spectrum of the U.S. population (CDC 2021a; DEA 2018; Gladden et al., 2019).

Recent commentaries suggest direct relationships between COVID-19 and addiction, specifically rates and treatment outcomes (Leppla and Gross 2020; Lister et al., 2020, US Census Bureau 2021); however, there are few empirical studies showing these relationships in concrete detail. Among the handful that have been recently published, Niles et al. (2021) analyzed national clinical drug testing data before and during the pandemic and documented that while rates for high-risk drugs such as non-prescription fentanyl and heroin had increased (i.e., 35%: 44% respectively), important geographic disparities in overdose and hospitalization exist that vary by the timing of stay-at-home orders and the differing COVID-19 impacts across each state (Niles et al., 2021).

One of the factors that has helped reduce overdose mortality rates has been expanded access and use of naloxone by first responders and other health care providers. Naloxone is an opioid antagonist that binds to opioid receptors to help prevent overdose mortality. Within the state of Pennsylvania, Act 139 was approved in 2014, which allows first responders including law enforcement, fire fighters, EMS, or other individuals to administer naloxone. In a localized study of Marion County, Indiana, Glober et al. (2020) observed an increase in suspected overdose EMS service calls (43%), EMS naloxone administration (61%), and drug overdose deaths (47%) following the Indiana stay-at-home order, with the most pronounced increases concentrated in spatially high-population density zip codes (Glober et al., 2020). This study suggests that since naloxone is often prescribed upon discharge from the hospital and can be readily purchased at pharmacies, those with better access to pharmacies are more likely to acquire and use naloxone in the case of an overdose, and especially during COVID-19, these individuals have better chances of survival. These emerging trends suggest granular attention is needed on the spatially varied impacts of COVID-19 on the opioid epidemic.

This attention is especially critical given that the interactions between COVID-19 and addiction patterns are likely to vary geographically based on histories of deindustrialization, agrarian and energy economies, geographic isolation, and the strength of social networks that can support those struggling with pain and addiction (Wang et al. 2013; Frank et al., 2016, Dasgupta et al., 2018). Current literature demonstrates important geographic disparities in overdose, overdose response, and healthcare access for populations using opioids (SAMSHA 2021, Salamina et al., 2010; Amato et al., 2005; Prunuske et al., 2014). Both state-level and national-level data suggest greater rates of opioid prescribing and opioid overdose in rural areas in contrast to urban localities (Hedegaard and Spencer 2021; Keyes et al., 2014; García et al., 2019). The ability of people to access and utilize treatment options, such as naloxone, clinics, and hospitals across the United States often follows an inverse relationship whereby individuals most in need of requisite health services live greater distances from healthcare, and must spend longer times travelling to facilities, varying by rural, peri-urban and urban classification (Abraham et al., 2018; Luu et al., 2019; Prunuske et al., 2014).

Yet even given these existing studies, the main factors are often more nuanced and vary at the county scale. As Lippold and Ali (2020) reveal in their analysis of opioid-involved overdose deaths across metropolitan and non-metropolitan areas of the U.S., significant heterogeneity in the rates of opioid-involved overdose deaths exist within and between metropolitan and non-metropolitan areas, supporting the existence of sub-epidemics in the ongoing national opioid crisis (Lippold and Ali 2020). Additionally, Lister et al. (2020) underscore the value of county level examination to elucidate the often understudied urban-rural dynamics undergirding the opioid epidemic. By examining annual changes in opioid related outcomes between 2013 and 2017 across Michigan, the authors reveal that while rural counties experienced higher opioid prescribing rates than urban counties, urban counties experienced higher opioid overdose death rates compared to rural counties, underlining the need for regional, country-level investigations to examine how and why trends may diverge or align with national patterns. As recent data from the National Vital Statistics System (NVSS) (2021) reveal, national overdose patterns are often inconsistent with rural-urban opioid use and overdose and treatment configurations at the county scale and are differentially experienced across the population (Hedegaard and Spencer 2021).

While no clear pattern in overdose rates has been demonstrated by socioeconomic status, people who are male, and those who are White or Native American have been consistently shown to overdose at higher rates (Altekruse et al., 2020; Marshall et al., 2019; Carpenter et al., 2017). Initial studies have also indicated higher overdose rates for middle-aged White men, compared to other age groups, and those with lower educational attainment (Case and Deaton 2015). Even with significant state trends, concern remains that rates of opioid use and misuse do not accurately capture the severity of the epidemic. As evidence of this, one study indicated that published rates of opioid overdose mortality were underestimated in several states, with Pennsylvania getting special mention for having one of the largest undercounts (Case and Deaton 2015). Finally, much has been made in the popular press of rural-urban differences in opioid use and mortality (Koh and Mostoller 2019; Del Real 2018). However, recent empirical studies examining these patterns have found a more nuanced picture; although opioid overdose mortality seems to be higher in rural areas, non-fatal overdose and opioid misuse rates appear to be more evenly distributed (Ruhm 2017; Pear et al., 2019). In other research from Pennsylvania, Holmes et al. (2022) find statistically significant variations in naloxone dispensation across urban and rural counties with direct impacts for overdose mortality. Since overdose mortality is a much rarer incident than overdose morbidity, opioid overdose patterns in general necessitate urgent attention.

Additionally, emerging research is pointing to patterns of nonmedical opioid use (NMO) in both urban and non-urban environments with particular attention to the role of the built and social environments. As Tempalski et al. (2022: 701) explain, “important gaps in the scientific literature currently limit our understanding of how both physical and social features of environments shape risk for NMO overdose in rural and suburban settings and therefore limit our ability to intervene effectively.” In order to address this limitation, the authors propose a framework that addresses socio-built environments, which are constituted of environmental characteristics across both urban and non-urban settings, in addition to the social features of built environments. Their study is effective in broadening analyses of opioid use and misuse to consider generalizable, as well as place-specific, factors that influence the effectiveness of treatment responses.

Given the ways that the COVID-19 pandemic is potentially intensifying the factors that contribute to substance misuse, research on the socio-geographical variations in opioid misuse would assist in addressing place variations in substance use, addiction patterns, and treatment responses. As Watson et al. (2022; 1) argue, although “the decades-long opioid epidemic and the more recent COVID-19 pandemic are two interacting events with significant public health impacts for people with opioid use disorder (OUD) … most published studies regarding the intersection of these two public health crises have focused on community, state, or national trends using pre-existing data.” The consequence is that the place-specific factors underpinning these variations, along with the ways opioids are differentially experienced among the population, are underemphasized in the existing literature. Indeed, they conclude that “there is a need for complementary qualitative research aimed at identifying how people with opioid use disorder (OUD) are understanding, experiencing, and navigating this unprecedented time” (Watson et al., 2022; 1). Equally, more work that considers the differential needs, experiences, and realities of OUD health providers are needed to explicate the social and structural drivers underpinning geographic variations to ensure opioid users have timely access to the full continuum of evidence-based interventions (Marchand et al., 2022; McCann et al., 2022).

Like any other epidemic, the substance addiction crisis in the United States has a wide range of interconnected geographic dimensions. Victims of this epidemic are part of a complex network of locations that include their family and social networks, communities, health service areas, drug sourcing locations, treatment centers, and legal/policy jurisdictions. Rural areas have disproportionate shares of at-risk groups, including American Indians, military veterans, older adults, people with disabilities, and people living in poverty (Rigg et al., 2018), and recent research identified each of these population characteristics to be associated with higher county-level drug mortality rates (Monnat 2018). Deindustrialization within rural America, twenty-first century energy economies that produce greater economic insecurity, social isolation, and the outsourcing of manufacturing, generate high rates of unemployment, injury, and chronic pain that are differentially experienced by social groups (Dasgupta, Beletsky and Ciccarone 2018; McLean 2016). Reflecting this, the National Academy of Sciences, Engineering, and Medicine (2017: 41) emphasizes that “While increased opioid prescribing for chronic pain has been a vector of the opioid epidemic, researchers agree that such structural factors as lack of economic opportunity, poor working conditions, and eroded social capital in depressed communities, accompanied by hopelessness and despair, are root causes of the misuse of opioids and other substances.”

As the next section details, qualitative interviews were conducted with partnering public health providers located primarily in Southwestern Pennsylvania where the epidemic has been particularly severe. Southwestern Pennsylvania experienced an economic boom in the 1940s due to steel production in support of the war effort. Labor concessions in the 1950s and international competition made steel production untenable, and the closure of the Donora mill in 1967 precipitated the economic and population decline that would spread throughout the region. By the 1980s, steel capacity utilization fell nearly 30 percent prompting economic decline and population outmigration (Guza 2014). Southwestern Pennsylvania is a former steel region that flourished because of its proximity to the city of Pittsburgh, but also due to its proximity to the Monongahela River for the transport of materials and wide availability of coal for producing coke for steel production. These interacting economic and environmental conditions generated the boom for towns in Southwest Pennsylvania during the 1940s and 1950s, and subsequent industrial decline, which is credited for contributing to outmigration, economic decline, and increasing crime that are tied to the opioid epidemic (Guza 2014). As the COVID-19 pandemic continues to inflict uneven realities for opioid users OUD health providers, research and policy attention to opioid use and misuse remains urgently needed (Glober et al., 2020; McCann et al., 2022; Rosenbaum et al., 2021; Watson et al., 2022).

3. Case study and methods

Pennsylvania has been among the states with the highest rates of mortality due to drug overdose in recent years, with a majority of these attributable to opioids (NIDA 2020). From 2010 to 2019, rates of opioid-related mortality per 100,000 population in Pennsylvania rose from 5 to 23.7, an increase of nearly 475% (CDC 2020). The CDC reported 3034 deaths due to opioids in 2019, up from 2866 in the previous year (CDC 2020). Heroin and fentanyl are the largest threats to overdose in the state, exacerbated by a recent surge in illicitly manufactured fentanyl and fentanyl-related substances (DEA 2018). Increased fentanyl availability specifically contributed to a 65% spike in overall drug overdose deaths in Pennsylvania between 2015 and 2017 (DEA 2018).

Despite these trends, state efforts to curb opioid prescribing and increase naloxone distribution prior to the COVID-19 pandemic had effectively decreased drug availability and reduced mortality rates (DEA 2018). These policy measures included implementation of limits on first-time opioid prescriptions, a Prescription Drug Monitoring Program, and a standing order allowing naloxone dispensation without a prescription (Pennsylvania Department of Health 2021). Some of these responses extended to other states, and in 2018, there was a decline in the national drug overdose death rate for the first time in 28 years, mostly attributable to fewer opioid overdoses (CDC 2022). Subsequently, life expectancy increased after a troubling multiyear decline (Hedegaard et al. 2020). However, the COVID-19 pandemic has potentially reversed these gains, either by intensifying patterns of opioid misuse and overdose or by transforming the places and populations most directly impacted by the opioid epidemic.

Given previous patterns of overdose incidence, coupled with differential survival rates, research is needed to understand the full extent of the COVID-19 pandemic for patterns of opioid use and misuse. While initial estimates of opioid overdose prevalence and mortality during the pandemic have been published, with the exception of the CDC's recent national estimates, these have largely focused on the experiences of particular cities or health care systems during the first half of 2020 (Glober et al., 2020). Comparatively, the largest spike in U.S. coronavirus cases and mortality occurred between November and February of 2020, with an additional upward trend occurring in April 2021 (CDC 2021b). The CDC also concentrates on overdose mortality, which does not capture the full expanse of the opioid epidemic. Existing research is demonstrating that with expanded access to naloxone, opioid morbidity represents a significant component of overdose incidents (Scholl et al., 2018; Zibbell et al., 2019; Holmes et al., 2022). Existing studies also focus more on overdose rates and trends during the pandemic with less attention to the socio-geographic correlates of these trends, or the lived realities of OUD providers during the pandemic (Glober et al., 2020; Alter and Yeager 2020; Rosenbaum et al., 2021; Watson et al., 2022). Thus, little is yet known about how these trends play out across disparate geographies, how overdose rates in 2020 compared to years prior to the pandemic, and how existing social and economic vulnerabilities may have combined with the trauma of the pandemic to transform the U.S. opioid epidemic.

3.1. Study sample

A mixed method approach was used to capture the complexity of opioid overdose and COVID interactions across and within the state of Pennsylvania (Curry and Nunez-Smith, 2015; Sammons, 2010). The quantitative component relied on data from the Pennsylvania State Police's Overdose Information Network (ODIN) and Department of Health's (DOH) dataset for doses of naloxone administered by Emergency Medical Services (EMS) (Pennsylvania State Police 2020; Pennsylvania Department of Health 2020). The ODIN dataset provides information on incidents of fatal and non-fatal drug-related overdose and response reported by PA criminal justice agencies and some third-party first responders (i.e., EMS, fire departments, medical staff, etc.). Research being conducted as part of a larger project has shown that Pennsylvania is unique in having this type of publicly available dataset on opioid overdoses. For example, West Virginia and Kentucky are two states with similarly high historical patterns of opioid misuse; however, neither state has an equivalent dataset. This presents an ideal opportunity to examine how the convergence of these two health crises, particularly during the first year of the COVID-19 pandemic. While the ODIN dataset offers the most up-to-date and comprehensive data on incidents of opioid-related overdose available at the state level, it is based on a voluntary system of reporting and therefore does not represent all overdose incidents involving criminal justice agencies or first responders within the state (Pennsylvania State Police 2020).

Analyses of opioid-related overdoses were conducted from 2018 to 2020. Opioid-related overdose incidents were defined as any overdose involving buprenorphine, carfentanil, fentanyl, fentanyl analogs or other synthetic opioids, heroin, methadone, or pharmaceutical opioids, based on the categories of reporting in the ODIN.

Age-adjusted rates of opioid-related overdose incidents, per 100,000 population, were calculated by county and year for 2018, 2019 and 2020. The 2000 county population was used as the standard for age adjustment. Unadjusted rates were also calculated by demographic group. The ODIN dataset contained duplicate records for some incidents as a result of multiple responders recording information into the dataset There were also a few incidents that involved more than one individual; however, these counted as one overdose incident. After systematically identifying and removing true duplicate records there were 15,719 unique entries from an original dataset of 26,542 records. The percentage change in age-adjusted overdose rates from 2018 to 2020 was calculated for each county. Statistical significance of change was determined using Poisson rate tests by county, which measure statistical differences in the rate distribution before and during COVID with a 95% confidence level.

To examine how COVID-19 affected OUD health and social care in communities, we complemented these quantitative analyses with in-depth interviews with opioid health providers via Zoom (video optional). The interviews with opioid health providers are part of a larger project examining the causes and consequences of opioid use in the broader Appalachia region including the states of Pennsylvania, West Virginia, and Kentucky. For this study, we purposefully draw on five in depth interviews with health providers from Pennsylvania who are partners in this project to help contextualize the quantitative data analyses. In order to acquire a range of perspectives and experiences, multiple recruitment strategies including purposeful sampling along with supplementary snowball sampling was used to recruit participants (Heckathorn, 2011). Individuals could participate in the study if they worked in OUD health and social care, were employed in Pennsylvania, and could speak to the issues affecting individuals managing addiction. Participants were selected based on their role and ability to inform policy within their organization as well as their ability to speak to a range of COVID-19 – opioid use interactions. This offered the capacity to identify commonalities and variations in organizational experiences and responses across counties (Creswell et al., 2003).

Prior to conducting the interview, verbal consent from all participants was received. All collected information was encrypted, and security stored without personal identifiers to protect confidentiality. Ethical approval from Penn State University was also received. All digital recordings were transcribed verbatim and proofread by all authors. Integration of both quantitative and qualitative data occurred in three ways; at the level of design, method and interpretation and reporting of results (Fetters et al., 2013). At the design level, an embedded convergent parallel design was used; at the methods level, building and merging was employed, and at the interpretation and reporting levels, data was integrated narratively (Fetters et al., 2013). This provided the capacity to compare and triangulate findings and illuminate multiple perspectives and ecological levels or processes and outcomes (Ivankova and Kawamura, 2010).

4. Results

The results of the analysis demonstrate statistically significant county-level changes in the age-adjusted rate of opioid-related overdose incidents before and after the onset of the COVID-19 pandemic in Pennsylvania, which were bolstered in the qualitative interviews. Fig. 1 shows the significant change in overdose rates from 2019 to 2020 with increases represented in red and decreases represented in blue.

Fig. 1.

Significant changes in opioid overdose rates in Pennsylvania by county from 2019 to 2020.

A total of 19 counties experienced a significant increase in opioid overdose rates in 2020, with Allegheny, Bradford, Erie, Forest, Somerset, Susquehanna, Venango and York showing the most significant increases (p < .001). These counties range in population size (1,216,045; 60,323; 269,728; 7247; 73,447; 40,328; 50,668; 449,058 respectively), health insurance coverage (95%; 91.8%; 92.9%; 92.8%; 92.4%; 91.5%; 93.1%; 93.9% respectively), poverty rates (11.3%; 12.6%; 15.5%; 16.7%; 10.7%; 11.7%; 13.5%; 8.7% respectively), and rates of violent crime (394.5 per 100,000; 181.2 per 100,000; 224.8 per 100,000; 468.5 per 100,000; 114.8 per 100,000; 126.1 per 100,000; 134.6 per 100,000; 219.9% respectively) (See Healthiest Communities, 2022a, b, c, d, e, f, g, h, i). Meanwhile, five counties saw significant decreases in incident rates, with Mercer, Potter and Wyoming Counties showing the most significant decreases (p < .001). Counties that saw significant decreases equally range in population size (109,424; 16,526; 26,794 respectively), health insurance coverage (92.9%; 92.7%; 94% respectively) and unemployment rates (9.7%; 9.4%; 8.4% respectively) (See Healthy Communities, 2022j, k, l). Several OUD providers elaborated on these socio-geographic patterns of opioid overdose during the first year of the COVID-19 pandemic, articulating how some place-specific factors such as violence and crime, along with variations across and between the Pennsylvania counties, were influencing patterns of substance misuse. Participants explained that although factors underpinning county variations in overdose differed, countries across the state were experiencing similarly high rates of overdose. As explained in the following excerpt by a provider who serves both urban and rural localities, while opioid use is increasing across the state, different factors tied to urban and rural geographies are associated with the increase in overdose:

“Western Pennsylvania, where I'm from, it's, there's a lot of variation between counties. So, you can drive 20 min across the line, and it almost seems like a different state. So, it's absolutely true. I think the rates of use are similar, but you see sort of different sort of associated factors with substance use in these counties. Like in Fayette County, there's a lot of violence associated with drug use. And in Greene County, there really isn't much, you know. It's more of a rural county, there's more land. So, there's definitely some differences, but I think the use rates are pretty similar.”

Fig. 2 shows the age-adjusted opioid overdose rate from 2018 to 2020. The median overdose rate increased from 34/100,000 in 2018 to 37/100,000 in 2019 and 40/100,000 in 2020. The mean rate was higher in 2018 (45/100,000) and again in 2020 (49/100,000) compared to 2019 when the interquartile range was also the narrowest. In 2018 and 2019, the highest rates were in Dauphin County, home to the state capital of Harrisburg, which was surpassed in 2020 by York County. This is also depicted in the change shown in these counties in Fig. 1.

Fig. 2.

Pennsylvania age-adjusted opioid overdose rate by year, ODIN 2018–2020.

Interviews with addiction treatment providers helped contextualize these increases. One interview conducted with an addiction treatment specialist explained how the onset of the COVID-19 pandemic and associated stay at home orders presented significant psychosocial challenges for clients and their ability to manage their personal health. Common across discussions were remarks of “increases in depression and anxiety” and how these issues fueled additional pressures for those seeking addiction treatment services. As one substance use provider located in an urban center below explains, pre-existing social dynamics related to inequities in transportation, employment, housing and social relations influenced the ways that the COVID-19 pandemic is intensifying and potentially transforming the U.S. opioid epidemic with similar, yet unique impacts depending on one's sociogeographic location:

“No question that the social determinants of health have really either exacerbated a person's substance use, or their substance use coupled with the pandemic has exacerbated the social determinants of health or made certain things worse. People already maybe had say, transportation issues. Those were sort of magnified during this time. People had unemployment or underemployment. Certainly, that's been magnified during this time. Stress in terms of strained relationships with loved ones. Everybody's stressed out. That's gotten worse. Finances, you know, or lack thereof, that's gotten worse. So, all of these things that can kind of contribute to a person having a low socioeconomic status or having just multiple social determinants of health impact negatively, that's all really, really gotten worse. And so, that's one thing. I think people that have struggled in those areas have really had a difficult time staying in recovery during the pandemic.”

To explore how COVID-19 impacted emergency responses and mortality in Pennsylvania, we examined the age-adjusted opioid overdose mortality rate from 2018 to 2020. As indicated in Fig. 3 , there has been a steady increase in mortality from overdoses, starting at a median rate of 25/100,000 in 2018 and rising to 29/100,000 in 2020, even with greater provision of naloxone and other emergency responses. Similar to overall overdose rates, Dauphin County, a county located in mid-east Pennsylvania, with a population of approximately 286,401 people, which constitutes a higher population density per square mile, higher crime rates, and lower life expectancy compared to the state averages (i.e., 530 versus; 286; 100 arrests per 1,000 residents versus 75 arrests; 82.4 years versus 83 years) (Pennsylvania Department of Health, 2020; Berkely Education, 2020), had the highest opioid overdose mortality rate in the state in 2018 and 2019. Yet in 2020 York County, a county known for manufacturing and located in southeastern Pennsylvania bordering Maryland, with a population of 447,628 people, a higher percentage of white and Hispanic populations, and a higher hospital average inpatient occupancy rate, had the highest rate of opioid overdose mortality (Pennsylvania Department of Health, 2019).

Fig. 3.

Pennsylvania Age-adjusted opioid overdose mortality rate by year, ODIN 2018–2020.

Although the state of Pennsylvania enacted the PA Act 139 in 2014, which allows first responders to administer naloxone to people with suspected overdoses and provides immunity to those responding to and reporting overdoses, OUD providers highlighted continued inequities in the administration and provision of naloxone in the region that was exacerbated during the pandemic. Participants highlighted issues ranging from a lack of acceptance of harm reduction approaches, particularly acute in rural areas, limited funding for harm reduction strategies like naloxone provision, along with stigma and tenuous relations between police and OUD clientele. Underpinning their comments was a general sense that simply allowing police and responders to administer naloxone was insufficient, without addressing the structural harms, over policing and entrenched stereotypes that characterize harm reduction approaches and relations between opioid users, communities, and opioid responders. As highlighted below from an informant serving urban and rural populations, intersecting dynamics associated with stigma, geographic disparities and isolation are colliding with COVID-19 leaving providers struggling to effectively support their clientele who are managing active addiction:

“Uh, some of them won't ask [for Narcan]. They'll only ask when they're not doing on their own accord. But when we talk to a client, and I know they're using, the conversation will go like, listen, if you're using, I'm just going to give you Narcan, you know. You should keep some on you. Are you using? Yeah, I am. Okay. You know, like we're having to like get that out of them. But that's how that works. A lot of them won't ask because a lot of people are scared, the whole COVID as it is. Thank goodness for telehealth because like that has filled in gaps. Those people that are, that really need a Narcan, that really need naloxone, like that are living in a house that is deplorable conditions, that is using and really need that. We still do go out, but it's a whole technique now. You know, it's, we have to be very safe, of course. And, you know, some people, we might have to throw ‘em Narcan from our car. Like it's hard, you know, and we go in like some dangerous areas, but that's how it's done now. Like the police can give them [Narcan]. There's more compassion now, but stigma, it's still there. Especially now with COVID.”

To further unpack population variations, we analyzed how overdose was differentially experienced among men and women. Fig. 4 shows the differences in overdose rates among men and women from 2018 to 2020, both demonstrating a rate decline from 2018 to 2019 followed by a jump in 2020 and an upward trend.

Fig. 4.

Pennsylvania opioid overdose rate by sex, ODIN 2018–2020.

OUD providers offered context to these temporal gender variations. Although several indicated gender variations in overdose prior to the pandemic, they remarked that once the pandemic began, they noticed similar overdose rates across gendered lines. As one participant located in an urban setting, yet travels across the state to help those in need below explains, although men and women have experienced different challenges during the pandemic, their introductions to drug use can occur in different ways:

“So, I can tell you that if a woman comes in because she's got a drug problem, it's very likely that she was first introduced to whatever it is that she's using by a male. The boys are just more likely to experiment and to try things, and then they are more likely to show the girls what they're doing and how good it feels. So, they were most often first time using something, and that was introduced to them by, you know, a male.”

Fig. 5 illustrates overdose rates from 2018 to 2020 among Black and White individuals. The trends closely mirror each other with slight decreases in rates from 2018 to 2019 and then sizeable upticks from 2019 to 2020. In keeping with recent literature, the overdose rate among Black Pennsylvanians is higher than that among White individuals.

Fig. 5.

Pennsylvania opioid overdose rate by race, ODIN 2018–2020.

OUD health providers articulated the historical and place specific factors underpinning these racial variations. Participants explained that although the Black population comprises only a small proportion of the total population in the state, they have begun witnessing an uneven increase in overdose. They explained this uptick was due to a paucity of Black health providers and racial discrimination in the health system, rooted in broader systems of structural racism and racial capitalism in the country (Garcia, 2022; Laster Pirtle, 2020). As the participant below explains, histories of oppression together with contemporary forms of structural racism, lay the foundation that created the present environment of medical distrust, stigma, and health care avoidance:

"I think one, the barriers to care. So, presence of African American providers in the communities where African Americans live, there's really a paucity of providers. So that's the first problem that creates some mistrust. Historical injustices like the Tuskegee syphilis study. And coupled with historical disparities in terms of treatment for medical conditions overall, how many of our people in this country get heart catheterizations, for example, or, you know, invasive procedures, or high-tech new treatments for conditions. You look at that, there's disparity, which creates more mistrust. So, I think those are some of the barriers really. Some are financial, and then some are just stigma in terms of historical thoughts about healthcare in the African American community. So, you know the Black Church plays a very pivotal and important role in terms of, you know, what we do as a community. And, you know, because of that, it's interesting, a lot of people might seek first the counsel of their pastor or local important people in their own communities and not necessarily think, think about health care. But I think really all those things, you know, really play, play into why this happens."

Taken together, the quantitative and qualitative data reveal complicated relationships between the COVID-19 pandemic and the U.S. opioid epidemic. The spatial analyses demonstrate geographically distinct changes in opioid overdose rates across counties during the first year of the COVID-19 pandemic, in addition to variations across populations. The interviews with public health providers and addiction treatment counselors reveal some of the ways that the pandemic has been differentially experienced by providers as well as those dealing with addiction.

5. Discussion

By integrating spatial-temporal data with the lived realities of health providers in places across Pennsylvania, this study demonstrates key variations in county level rates of opioid overdose prior to, and during, the COVID-19 pandemic. Foremost, the statistically significant differences in county-level incident rates suggest that many counties experienced worsening overdose trends in 2020, while a handful of others did not. As detailed in Fig. 1, 19 counties experienced a statistically significant increase in rates of overdose from 2019 to 2020. The significant increase within specific counties underscores important geographic variations in the ways COVID-19 policy responses likely intersect with place-specific factors – changing agrarian economies, growing housing instability, disconnected health services, and fraying social support networks – embedded in parts of Pennsylvania that could support those with OUD (Pennsylvania State Police 2020; Pennsylvania Department of Health 2020).

As underscored by health providers, these factors are likely compounded by initial stay-at-home orders that brought unique challenges for OUD care, such as interrupted treatment, limited opportunities for safer drug use practices, and fewer drug dispensing options, and heightened psychosocial trauma (CDC 2021a; 2021c). Since many treatment centers have been forced to close or significantly scale back during the pandemic, opioid users are left with limited social support and access to vital health services necessary to support recovery and treatment (Pennsylvania Commission on Crime and Delinquency, 2020). Furthermore, the significant increases in overdose among racialized populations highlights changing racial compositions of opioid use in Pennsylvania and underscores the need for critical attention to the intersections of race, structural inequities and harm reduction in the region (Lopez et al., 2022).

This study equally uncovers important declines in opioid overdose rates from 2019 to 2020. Despite the challenges of opioid care provisioning following stay-at-home orders, our results demonstrate that some counties experienced significant declines in opioid overdose rates that are consistent with previous state trends in 2017 (NIDA, 2020). In contrast to rising rates of opioid overdose throughout most of the state, declines in particular counties may be due in part to the opioid disaster declaration and associated public health responses that aim to reduce the risk of substance misuse through education, limiting the availability of addictive substances and combatting stigma, which may lead to a reduction in the number of overdose incidents reported to police or emergency medical services ( Alexander et al., 2020). Yet, others suggest that declines could be due to changing drug trafficking patterns associated with disrupted travel and international physical distancing guidelines (UNODC 2021 ). Since Interstate 95, recognized as the East Coast's major drug transportation corridor (National Drug Intelligence Center 2006), runs directly through counties in Pennsylvania's south-eastern corner, providing drug access to many midsize and smaller towns in this region, it is possible that changing transportation patterns during the COVID-19 pandemic may be contributing to declining opioid overdose rates in these counties. Such geographic variations suggest the need for greater attention to the place-specific factors and determinants of opioid overdose at the county level in the wake of the COVID-19 pandemic.

Moreover, since first responder organizations in need of naloxone must first reach out to their relevant CCE directly to request naloxone from the state-wide allocation (Pennsylvania State Police 2020), bureaucratic processes related to short staffing and underfunding, exacerbated during the pandemic, likely compound challenges of naloxone acquisition and undermine access for needy populations (Pennsylvania Department of Health 2020). As the qualitative interviews indicate, stigma, geographic resource disparities and political tensions continue to make the administration and provision of life saving support particularly challenging, especially during pandemic. While Pennsylvania's first responder program launched in 2020 aims to increase the availability of life-saving medication for individuals at greatest risk of opioid overdose (Pennsylvania Department of Health 2020), the results show the program has not fully realized its goal in providing equitable access to naloxone treatment, especially in the wake of the pandemic (Holmes et al., 2022).

There are limitations to this study that warrant mention. First, we relied on data from the Pennsylvania ODIN dataset, which is maintained by the Pennsylvania State Police and accounts for overdose incidents to which police and other emergency responders were called. More than 600 agencies in Pennsylvania contribute to this database, but it is a voluntary reporting system and may not include entries from all emergency responders in the state, nor does it incorporate bystander intervention in cases when administrative agencies did not respond. We cannot estimate whether there are systematic differences in which agencies report incidents by county, which may create bias in county-level comparisons. However, the relative difference in incidents between counties and the number of incidents represented provide useful comparisons for policymakers and researchers. Second, since this was a cross-sectional study, we can only provide a snapshot of the frequency in opioid overdose at a given point of time, and thus there is a limitation in determining whether the increase in opioid overdose followed directly from factors or policies related to the pandemic. Third, opioid overdose data was only available at the county level, thus limiting our ability to identify geographic variations in rates and incidents of opioid overdose at the block and/or Census tract level. While the qualitative interviews do provide insightful contextual accounts of COVID-19-opioid use interactions in place, they are not representative of health providers across the state. Despite these limitations, this study provides novel and timely insight on the potential interactions between opioid use and COVID-19 policy responses and impacts, transferable to other Appalachian states. Furthermore, given the voluntary nature of the ODIN dataset, it is possible for additional reporting in the context of COVID-19.

6. Conclusion

Our study suggests that state efforts to mitigate COVID-19 in Pennsylvania can have unintended consequences for opioid overdose and responses that vary across spatial and temporal scales. Although government measures have been helpful to support public health, inequities in their implementation have resulted in missed opportunities to engage the lived realities and entrenched disparities that underpin Pennsylvania's opioid problem. Attending to persistent issues of economic stress, access disparities, along with stigma and entrenched processes of racism would serve as a foundation for longer-term changes that make health and social care more accessible and result in a nimbler system that can respond faster during the next public health crisis (Green et al., 2020; Becker et al., 2021; Nunes et al., 2021)

The results of this study should help guide future research and new targeted policy interventions that aim to expand naloxone administration, extend substance use treatment facilities and mitigate the negative implications of COVID-19 for the most high-risk areas for addiction and opioid abuse. For instance, as more services shift towards online platforms, there remains a continuous need for low-barrier, flexible and in-person services and appointments to consider the differential needs of individuals who do not have consistent access to phones or internet, who are precariously housed, or street involved (McCann et al., 2022).

In particular, our findings demonstrate a need for more informed policy interventions that attend to the differential experiences and place-based realities of OUD providers and those struggling with addiction. Ensuring funds are appropriately allocated to address the effects of social isolation and broader social and structural inequities would not only help mitigate relapse and overdose during the pandemic but provide a foundation on which to expand care provision (Silva and Kelly, 2020; Watson et al., 2022). As the COVID-19 pandemic continues to unfold in new and uncertain ways, attention to the contextual realities of people with OUD is essential to avoid unintended consequences of well-meaning policies, such as the rise in overdose following the 2020's shelter in place orders and social distancing measures (Glober et al., 2020; Rodda et al., 2020; Slavova et al., 2020; Rosenbaum et al., 2021). Ultimately, interactions between pre-existing place-based inequities, colliding public health crises and associated responses are in need of urgent consideration to improve public health.

Data availability

Data will be made available on request.