The impact of the novel coronavirus disease (COVID-19) pandemic on drug overdose-related deaths in the United States and Canada: a systematic review of observational studies and analysis of public health surveillance data

Drug overdose-related deaths hereunder were operationalized as drug poisoning deaths attributable to, but not limited to, opioids, cocaine and amphetamines, which encompassed different types of drugs (pharmaceutical, non-pharmaceutical [i.e. illegal]), places of death (residence, public place, unknown) and manners of death (unintentional, intentional, undetermined), as available.

The reporting of the systematic review was consistent with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Guidelines (Prospero Protocol Registration: CRD42021230223; See reporting checklist in Table S1 in Additional File 1) [15]. Electronic searches were conducted in Medline, Embase and PsycInfo to identify studies published from database inception to December 18, 2020 (see Additional File 1 for the search strategies). In order to be as comprehensive as possible, and to reduce the risk of publication bias, additional sources were drawn upon to identify studies: [1] keyword searches in Google Scholar were conducted [2], an issue brief listing reports of drug overdose-related deaths in the United States was reviewed [16], [3] a bibliography of published studies on addiction-related topics and COVID-19 was assessed [17], [4] reference lists of included studies were examined and [5] expert consultations were held [18]. The titles and abstracts were reviewed in the first round of screening, and the full texts were reviewed in the second round of screening. The selection criteria were as follows:

No language restrictions were applied, and unpublished literature was eligible for inclusion. Unpublished literature was included to capture all available evidence, as peer-reviewed manuscripts were expected to be limited in number due the timeliness of the topic. Prior reviews, animal studies, newspaper articles and opinion pieces (i.e. lacking empirical data; e.g. commentaries and research letters) were excluded. Similarly, routine public health surveillance reports were excluded (e.g. [19]), as they tended to include statistics on drug overdose-related deaths rather than quantify the impacts of COVID-19 on drug overdose-related deaths. However, data from them were extracted for the Percentage Change Analyses. Studies that met the selection criteria proceeded to the data extraction, where details on location, time period, population, outcome definition and ascertainment and main findings were noted. Risk of bias assessments were not conducted, as most studies were published as government documents or independent research reports.

Two authors (SI and JR) developed the electronic search strategies, which included keywords from the conceptual domains of mortality, drugs and COVID-19 (detailed search strategies for all databases are available in Additional File 1). Two authors (SI and FN) executed the electronic search strategies, screened titles and abstracts and reviewed the full texts. Disagreements between the two authors in the selection of studies were resolved by consultation with a third author (JR). One author (SI) completed the data extraction, with subsequent review and verification by another author (FN). Multiple reports from included studies were considered in the data abstraction.

As the outcome definitions were expected to vary considerably between the included studies, findings were not pooled quantitatively in a meta-analysis, rather findings were synthesized qualitatively in a narrative synthesis. The findings were grouped and reported according to two comparison periods for drug overdose-related deaths after the onset of COVID-19: 1) months leading up to the pandemic in 2020 (e.g. March to April 2020 vs. January to February 2020); and 2) comparative months from 2019 (e.g. March to June 2020 vs. March to June 2019). Percentage changes in drug overdose-related deaths from these studies were summarized as range of effects and directionality of effects (through vote counting). Findings from all pertinent analyses were additionally presented, as available.

Data from national or subnational (state and provincial) government public health surveillance systems and reports were compiled through electronic searches of Google (using keywords such as “drug deaths”. “opioid deaths” and “overdose deaths”), review of a listing of opioid dashboards and data assembled by the Carolina Center for Health Informatics [20] and expert consultations (with membership of the Canadian Research Initiative on Substance Misuse). These electronic searches for the percentage change analyses were not systematic, although a similar set of keywords were utilized. To be eligible for data extraction and analysis, data had to be available according to or be amenable to formatting as annual quarters (Quarter 1 [Q1]: January to March; Q2: April to June; Q3: July to September; Q4: October to December) minimally from January 2020 to June 2020. No exclusions were made on the basis of other criteria, including current reporting status (preliminary, finalized).

Data pertaining to the broadest available category of drug overdose-related deaths were extracted until Q3 2020. For example, drug overdose-related deaths were extracted rather than opioid overdose-related deaths, fentanyl overdose-related deaths or psychostimulant overdose-related deaths, if data on all four were available. If such a broad category of drug overdose-related deaths was not available, data pertaining to opioid overdose-related deaths were extracted, as they represent the main driver of drug overdose-related deaths in most jurisdictions in North America. Subnational data were extracted for 12 jurisdictions in the United States (Connecticut, Indiana, Louisiana, Maine, Massachusetts, Mississippi, New Hampshire, New Jersey, Rhode Island, Vermont, Virginia, Washington) and national data were extracted for Canada. However, as the national data were available until Q2 2020 in Canada, additional subnational data that extended until Q3 2020 were extracted from Alberta, British Columbia and Quebec. The primary analysis included computation of percentage changes in drug overdose-related deaths comparing Q2 2020 with Q1 2020 separately for all jurisdictions. The secondary analyses included: 1) computation of analogous percentage changes comparing the subsequent quarters to characterize the trajectory of the impact of COVID-19; and 2) computation of analogous percentage changes comparing quarters from 2020 with quarters from 2019 to contextualize the impact of COVID-19.

All research activities were conducted as part of the Ontario Node of the Canadian Research Initiative on Substance Misuse Project. No additional review and approval was required, as aggregated data were extracted from publicly available studies and public health surveillance systems and reports.

The findings from the percentage change analyses are presented in Table 2 (see Additional Files 2 and 3 for data sources, derivation notes and death counts). Jurisdictions in the United States reported variations of either drug overdose-related deaths or opioid overdose-related deaths. Drug overdose-related deaths increased by 2 to 60% in Q2 2020 compared with Q1 2020 in jurisdictions in the United States. Variations of either opioid overdose-related deaths or illicit drug overdose-related deaths were reported in jurisdictions in Canada. Drug overdose-related deaths increased 58% in Q2 2020 compared with Q1 2020 in Canada. During the same time period, drug overdose-related deaths more specifically increased 107% in Alberta, 80% in British Columbia and 28% in Quebec. Among jurisdictions that reported drug overdose-related deaths in Q3 2020, comparisons with Q2 2020 demonstrated increases in two jurisdictions in the United States and two jurisdictions in Canada (3 to 6% and 12 to 13%, respectively), and decreases in four jurisdictions in the United States and one jurisdiction in Canada (− 6% to − 30% and − 2%, respectively). On the other hand, drug overdose-related deaths in Q1 2020 compared with Q1 2019 demonstrated increases in all jurisdictions in the United States (1 to 72%), with the exception of Mississippi and New Hampshire. Decreases were evident in this regard in all examined jurisdictions in Canada (− 3% to − 9%). However, drug overdose-related deaths from Q1-Q2 2020 compared with Q1-Q2 2019 demonstrated increases in all but one jurisdiction in the United States and Canada (4 to 50%). This pattern of finding persisted when comparing Q1-Q3 2020 with Q1-Q3 2019 (2 to 62%).

Observational studies that examined the impact of COVID-19 on drug overdose-related deaths in the United States and Canada were reviewed. Drug overdose-related deaths after the onset of COVID-19 were higher compared with the months leading up to the pandemic in 2020 or comparative months in 2019. In additional percentage change analyses based on data from public health surveillance systems and reports, drug overdose-related deaths increased by 2 to 60% in jurisdictions in the United States and by 58% in Canada between Q1 2020 and Q2 2020. These findings indicate that drug overdose-related deaths increased after the onset of COVID-19.

Two rapid reviews have examined the impact of COVID-19 on drug overdose-related deaths based on empirical evidence [13, 14]. As they were both conducted during the early stages of COVID-19, there is minimal overlap with the studies or the public health surveillance systems data included hereunder [13, 14]. Largely on the basis of public health surveillance systems in Canada, increases in drug overdose-related deaths in British Columbia and Ontario (including Toronto and Simcoe Muskoka District Health Unit) were highlighted [13, 14]. However, it was difficult to conclude whether the increases were attributable to the toxic drug supply, COVID-19 or a combination of both [13, 14]. In a subsequent assessment from Philadelphia that was published after the completion of the systematic searches, Khatri and colleagues demonstrated increases in opioid overdose-related deaths between April 2020 and June 2020 compared with between December 2019 and February 2020 among non-Hispanic Blacks (mean monthly overdose deaths of 49 deaths vs. 30 deaths; p = 0.04), but not among non-Hispanic Whites (35 deaths vs. 45 deaths; p = 0.11) or Hispanics (14 deaths vs. 13 deaths; p = 0.57) [30].

Drug overdose-related deaths increased prior to and accelerated after the onset of COVID-19 in the United States. In contrast, drug overdose-related deaths stabilized prior to the onset of COVID-19, but increased after the onset of COVID-19 in Canada. Although seasonal variations (e.g. due to weather and holidays) may have contributed to some extent to the increase in drug overdose-related deaths, they probably do not explain the entirety of the findings. The percentage change in drug overdose-related deaths between Q1 and Q2 was 3% (mean; range − 40 to 38%) in 2019 and 21% (mean; range 2 to 60%) in 2020 in jurisdictions in the United States and − 4% in 2019 and 58% in 2020 in Canada, which suggests the potential role of the consequences of COVID-19. However, given the absence of studies with stronger methodological designs (e.g. interrupted time series design and regression discontinuity design), it is difficult to draw causal inferences because the impact of other factors cannot be ruled out. As such, these conclusions are preliminary that require further confirmation in future research.

Despite the preliminary nature of the current state of the evidence, it is clear that the rise in drug overdose-related deaths after the onset of COVID-19 necessitates actions on five fronts. The applicability of these recommendations will vary depending on local jurisdiction laws. First, expanded access to substance use disorder treatment needs to be facilitated, especially the delivery of services through innovative models of care to ensure minimal disruptions in treatment. Notably, telemedicine has facilitated low-barrier access to opioid pharmacotherapy with buprenorphine in the United States, as the federal regulatory changes enacted during the pandemic now permit clinicians to prescribe buprenorphine after conducting an initial visit using audio-visual or telephonic assessments rather than in-person assessments [31]. Removal of the in-person assessment requirement has resulted in rapid initiation of treatment, often including same day assessments, prescriptions and inductions [31, 32]. In addition, given the recent rise in psychostimulant overdose-related deaths [33, 34], treatment options for both opioid use disorder and psychostimulant use disorder are needed. Second, undisrupted access to harm reduction services needs to be ensured, including drug checking programs that make use of a wide range of technologies (e.g. fourier-transform infrared (FTIR) spectrometers, fentanyl test strips and drug testing kits), needle and syringe exchange programs and supervised consumption services, all of which have demonstrated varied range of benefits in terms of injecting risk behaviors, bloodborne infectious diseases and drug overdoses [35‐37]. Third, drug overdose prevention needs to emphasize risk reduction strategies during episodes of drug consumption: not to use drugs alone, determine the contents of drugs beforehand, ensure availability of naloxone on-hand and have someone check-in during consumption [29]. In particular, novel strategies such as virtual drug use spotting through social networks (e.g. phone calls, video calls) or technological innovations (e.g. smart phone applications) can be drawn upon when using drugs alone, an intervention where substances are consumed under observation and emergency services are alerted in the event of unresponsiveness [38, 39]. Wider availability of naloxone is similarly imperative in the current context, which has been shown to increase drug overdose emergency visits and decrease drug overdose deaths [40]. Fourth, to reduce drug use attributable risks and harms due to the toxic drug supply in North America, access needs to be provided to a safe drug supply (defined as legal and regulated supply of psychoactive substances that are typically available from the illegal drug market) [41, 42]. Fifth, the unprecedented circumstances suggest that there is a need for decriminalization of drug use, which will prioritize health over punishment, reduce stigma surrounding drug use and enable earlier engagement with prevention and treatment interventions [41, 43, 44]. Importantly, although these recommendations were relevant prior to the onset of COVID-19, their expedited, large-scale adoption is now high-priority given the rise in drug overdose-related deaths. In addition to these specific recommendations to address drug overdose-related deaths, prioritized access to vaccinations needs to be ensured for PWUD [45], who have intersecting social and health vulnerabilities that may elevate the risk for the infection, complications and mortality due to COVID-19 [46‐48]. PWUD experience homelessness and incarceration, high prevalence of pre-existing chronic health conditions, inadequate access to shelters and barriers to harm reduction services [8, 10, 49‐51]. In addition, substance use characteristics such as sharing paraphernalia and inability to social distance or self-isolate may further amplify the risks due to COVID-19 [49]. As such, access to vaccination is a priority for this population, which can be facilitated through the existing physical infrastructure of harm reduction and addiction treatment services [45].

There are some limitations that should be considered in the interpretation of the findings. First, non-fatal drug overdoses were not considered in the systematic review or percentage change analyses. As such, the full extent of the impact of COVID-19 on drug overdoses is probably larger. Second, despite the broad search strategies and expert consultations, studies from Canada were not well represented and studies from Mexico were not found in the systematic review, limiting the generalizability of the findings. In a similar manner, as the percentage change analyses included 12 jurisdictions from its states and territories, the findings may not be generalizable to the rest of the United States. Third, given the heterogenous definitions of drug overdose-related deaths in the included studies in the systematic review, quantitative synthesis of the findings through meta-analysis was not possible. Fourth, due to the need for timely evidence in the context of a rapidly escalating health crisis, data in some of the included studies in the systematic review were preliminary and subject to change, based on health surveillance systems or health administrative databases that are susceptible to misclassification bias and lacked confirmation through toxicological analyses or physician diagnoses. On that same note, data from all jurisdictions in the percentage change analyses were preliminary and subject to change. The onset of COVID-19 in particular has underscored the urgent need for standardized, real-time statistics on drug overdose-related deaths, as they are not available consistently across jurisdictions in North America. These limitations concerning the outcome assessment may have impacted the validity of the findings. Fifth, outcome assessment period in the included studies in the systematic review or percentage change analyses did not extend past September 2020, warranting a need for continued monitoring of drug overdose-related deaths. Sixth, stronger methodological designs were absent from most studies, and descriptive analyses (without statistical tests of differences) represented the analytical strategy in all but one of the included studies in the systematic review. Percentage change analyses were operationalized hereunder, as the granular data required for interventional time series analyses were not publicly available.