Real-world study of multiple naloxone administrations for opioid overdose reversal among emergency medical service providers

Abstract Background: The increasing rates of highly potent, illicit synthetic opioids (i.e., fentanyl) in the US is exacerbating the ongoing opioid epidemic. Multiple naloxone administrations (MNA) may be required to successfully reverse opioid overdoses. We conducted a real-world study to assess the rate of MNA for opioid overdose and identify factors associated with MNA. Methods: Data from the 2015–2020 National Emergency Medical Services Information System was examined to determine trends in events requiring MNA. Logistic regression analysis was performed to determine factors associated with MNA. Results: The percentage of individuals receiving MNA increased from 18.4% in 2015 to 28.4% in 2020. The odds of an event requiring MNA significantly increased by 11% annually. The adjusted odds ratio (aOR) for MNA were greatest among males, when advanced life support (ALS) was provided, and when the dispatch complaint indicated there was a drug poisoning event. Conclusions: The 54% increase in MNA since 2015 parallels the rise in overdose deaths attributable to synthetic opioids. This growth is visible in all regions of the country, including the West, where the prevalence of illicitly manufactured synthetic opioids is intensifying. Given this phenomenon, higher naloxone formulations may fulfill an unmet need in addressing the opioid overdose crisis.


Introduction
Despite substantial efforts by health care organizations, governmental agencies, harm reduction groups, and public health organizations to address the opioid epidemic in the US, the rate of opioid overdose deaths continues to rise exponentially and remains a significant public health concern. From 1999 to 2019, approximately 450,000 people died from overdoses involving any opioid, including prescription and illicit opioids. 1 Most recently, US deaths due to opioid overdose significantly increased 6.5% from 46,802 deaths in 2018 to 49,860 deaths in 2019. 2 Currently, the United States is experiencing a significant number of opioid overdoses fueled by an influx of illicitly manufactured fentanyl (IMF) and other high-potency synthetic opioids in pure or adulterated forms including carfentanil and other opioid analogues. 3,4 These synthetic opioids are considerably more powerful, with fentanyl and carfentanil estimated to be up to 100 and 10,000 times more potent than morphine, respectively. 5 Along with the higher potency compared to morphine, these synthetic opioids also have the ability to prolong respiratory depression even when present in only small amounts (e.g., 0.5 ng/mL). 6 While a prescription form of fentanyl has been available as a rapid-acting analgesic for pain management for some time, data indicates that the dramatic rise in drug overdose deaths in the past few years can be attributed primarily to the influx of illicitly manufactured synthetic opioids entering the market. 7,8 In 2018, there were 31,335 drug overdose deaths in the US specifically due to synthetic opioids, representing a rate of 9.9 per 100,000 population. In 2019, the rate increased to 11.4 per 100,000 population, with 4 out of 6 opioid-related deaths involving synthetic opioids. 9,10 Examination of data from 2013 to 2017 show the average annual increase in the rate of drug overdose deaths involving synthetic opioids was 75% per year. In contrast, this represents a marked increase compared to the average annual percent increase of 8.4% from 1999 to 2013. 11 The most recent published estimates indicate that overdose deaths due to synthetic opioids (primarily IMF) increased 38.4% from the 12-month period leading up to June 2019 compared with the 12-month period leading up to May 2020, suggesting that the problem remains unabated. 12 Furthermore, due to the disruption in treatment service delivery and harm reduction efforts caused by the COVID-19 pandemic, the situation has worsened; the average increase in the number of deaths (12months estimate) due to synthetic opioids increased by 185% from 770 pre-COVID-19 to 2,198 during the pandemic. 13 The regions in the US that have been impacted early on, specifically the Northeast, continue to suffer from the synthetic opioid crisis while the magnitude of the crisis is beginning to intensify in the Western region. In particular, the percentage of drug overdose deaths involving synthetic opioids was highest in the Northeast (71.0%) and lowest in the West (26.4%) in 2019. However, the increase in the rate of drug overdose deaths involving synthetic opioids was smallest in the Northeast while the increase in the rate was largest in the West. From 2018 to 2019, a 5.2% increase in the rate of drug overdose deaths was observed in the Northeast while a significantly large increase of 67.9% was observed in the West. 9 Altogether, recent assessment of the data indicates an expanding synthetic opioid overdose epidemic. 14 A key strategy for addressing the opioid epidemic, endorsed by the Centers for Disease Control and Prevention (CDC) and other medical organizations such as the American Medical Association (AMA), is expanding access to and administration of naloxone to reverse opioid overdose. Originally approved by the FDA in 1971, naloxone can be administered via intranasal (IN), intramuscular (IM), or intravenous (IV) routes by emergency medical service (EMS) providers or by trained bystanders (IN or IM administration). When administered in sufficient time and amount, naloxone is highly effective in reversing the symptoms of opioid overdose, including respiratory and central nervous system (CNS) depression. 15,16 Real-world evidence indicates that a percentage of opioid reversals require multiple naloxone administrations (MNA) for successful reversal, and that the frequency of MNA is increasing over time. [17][18][19][20] Notably, patients presenting in emergency departments (EDs) with overdoses involving IMF have been found to have excessively higher levels of drug concentration, requiring prolonged naloxone infusions, multiple bolus doses, or higher than standard naloxone doses for reversal. 18,[21][22][23] Using the National Emergency Medical Services Information System (NEMSIS) data, Faul et al. 19 indicated that the rate of MNA increased by 26%, from 14.5% in 2012 to 18.2% in 2015. More recently, Geiger et al. 20 observed a 43% increase in the rate of MNA, from 15.0% in 2013 to 21.4% in 2016.
While prior studies have shown significant increases in the rate of MNA and identified important factors associated with MNA, they do not reflect the current incidence of MNA given the last year of data examined by these studies was 2015 and 2016. 19,20,24 The increase in overdoses involving synthetic opioids since 2016 warrants an assessment of the degree to which the growth of MNA has continued into the present. Therefore, the objective of this study was to assess the rate of MNA administered by EMS from 2015 to 2020, as well as to identify the factors associated with MNA.

Data sources and sample
Publicly available data from the NEMSIS database from January 1, 2015 to June 30, 2020 was used for this retrospective analysis. NEMSIS is a voluntary database funded by the National Highway Traffic Safety Administration that includes data captured from EMS encounters from nearly all US states and territories, and is considered to be representative of the US population of EMS encounters. 25 NEMSIS data is based on individual event reports completed by EMS personnel and is submitted by local EMS agencies to the national database.
Data from 2015 and 2016 were part of version 2.2.1 of the NEMSIS database, while data from 2017 to 2020 were from version 3.4.0. It is noted that when transitioning from version 2 to version 3 of the database, the number of states contributing data to NEMSIS decreased from 49 states in 2016 to 35 states in 2017; correspondingly, the number of EMS events that were included also sharply decreased from 29,919,652 in 2016 to 7,907,829 in 2017. Therefore, the estimates calculated in 2017 (e.g., MNA rate) may deviate from the overall trend and should be interpreted within the context of a significant reduction in participating jurisdictions during 2017.
Notably, in 2018, the number of EMS events increased to 22,532,890 and in 2019 increased again to 34,203,087, which were more comparable to pre-2017 numbers.
For this analysis, the sample was restricted to all events where individuals received at least 1 administration of naloxone (where the medication given variable contained "naloxone," "Narcan," or "Evzio") between January 1, 2015 and June 30, 2020. Events in which opioid analgesics were administered or where the event was a medical transport or interfacility transfer were excluded. NEMSIS data does not contain codes for whether an incident involved an opioid overdose, therefore our sample of EMS naloxone administrations may also capture non-opioid-overdose incidents in which naloxone was administered as part of standard resuscitative procedures in emergency situations with an ambiguous underlying clinical cause. As the data in the NEMSIS database are de-identified, this study was exempt from review by the institutional review board.

Variables
The primary unit of analysis for this study was the rate of MNA, calculated by dividing the total number of individuals receiving more than one administration of naloxone by the total number of individuals receiving at least one dose of naloxone. The annual rate of MNA was calculated for the overall US as well as for each US Census Region for each year. Note that the rate of MNA for 2020 was based on 6 months (from January 1, 2020 to June 30, 2020) of data.
To identify factors associated with MNA, patient demographics (age and gender), clinical characteristics (complaint reported by dispatch and provision of oxygen), and operational elements (urbanicity, weekday/weekend, ambulance service level, incident/patient disposition, and destination) included in the NEMSIS data were assessed. Details of the specific categories of the age, disposition, and destination variables are provided in Appendix 1.

Statistical analysis
Patient demographics, clinical characteristics, and operational factors were summarized descriptively using "N" and proportion (%) for all variables across study cohorts by number of naloxone administrations for all years. The rate of MNA was assessed by year and by Census Regions. Confidence intervals (CI) were calculated to determine if the changes in the rate of MNA were significant between consecutive years. Unadjusted logistic regression models in which the binomial dependent variable was whether or not an event required MNA and the single independent variable included in the model was year (as a continuous variable) were used to determine if the odds of an event requiring MNA were significantly different each year across all, as well as within each, Census Region. Factors associated with MNA were assessed using a multivariable logistic regression model in which the binomial dependent variable was whether or not an event required MNA. The independent variables that were hypothesized to influence MNA decisions included demographic characteristics (age and gender), clinical factors (complaint reported by dispatch and provision of oxygen), and operational factors (urbanicity, weekday/weekend, ambulance service level, incident/patient disposition, and destination). The results of the multivariable logistic regression analysis were presented as adjusted odds ratio (aOR) with 95% CI. All data were analyzed using R version 4.0.0. 26

Results
From 2015 to 2020, of the 960,272 events where individuals received at least 1 administration of naloxone, 3,529 (0.4%) events were excluded because individuals were administered opioid analgesics during the event. Additionally, 9,746 (1%) events were excluded because they were classified as a medical transport or interfacility transfer, thus a final study sample of 946,997 events were included in the analysis ( Figure 1). Across all 6 years, over 80% of EMS-administered naloxone events occurred in urban areas, over three quarters of events occurred during the weekday, and mostly (>61%) among individuals who were male, between 20 and 64 years old (>80%), with approximately half residing in the south (Table 1). Approximately one third of the EMS events were handled by EMS personnel with BLS (Basic Life Support; generally refers to the type of care that first-responders, healthcare providers, and public safety providers provide to anyone experiencing cardiac arrest, respiratory distress, or an obstructed airway 27 ) service level. Over 85% of the individuals were treated and transported by EMS personnel to the hospital.
In all Census Regions, the rate of MNA significantly increased from 18.4% in 2015 to 28.4% in 2020 ( Figure 2). This represents a substantial increase in the rate of MNA by 54.3% which translates into an average annual growth rate of 9.1%. This continues the trend established by Faul (2017) and represents a 95.9% increase in MNA from 2012 to 2020 (see Appendix 2). Furthermore, the odds of an event requiring MNA increased by 1.11 times on average each year, which was statistically significant (Table 2).

Regional variation in MNA
While the overall rate of MNA displayed an upward trend from 2015 to 2020, there were substantial differences in the magnitude for different Census regions ( Figure 3). The South saw the largest increase in the rate of MNA of 106%, followed by 42.9% increase in the West, a 33.4% increase in the Midwest, and a 16.8% increase in the Northeast.
The average annual growth rate in MNA was 15.8%, 7.4%, 6.0%, and 3.1% in the South, West, Midwest, and Northeast, respectively. Likewise, the odds of an event requiring MNA significantly increased by 1.18 times in the South, 1.09 times in the West, 1.07 times in the Midwest, and 1.03 times in the Northeast on average each year ( Table 2).

Factors associated with MNA
A multivariable logistic regression analysis was also conducted to identify the demographic, clinical and operational factors that were most associated with MNA. Results from this analysis (Table 3) indicate that the odds of MNA were higher if the patient was male (aOR ¼ 1.15) and for individuals aged 20-29 and 30-39 (aOR ¼ 1.23 and 1.17, respectively).
Other factors that were associated with MNA were scene location, day of the week, dispatch information, and destination. The odds of an event requiring MNA differed by urbanicity, with the suburban and wilderness areas having the lowest odds of MNA (aOR ¼ 0.72, and 0.82, respectively) when compared to urban areas. The odds of MNA were also lower in rural areas compared to urban areas (aOR ¼ 0.88). An event occurring during the weekend had higher odds of MNA (aOR ¼ 1.06) compared to events occurring during weekdays. When the dispatch complaint (identified by emergency medical dispatcher) was identified as an ingestion/poisoning, the odds of MNA were higher (aOR ¼ 1.22) compared to when the dispatch complaint was identified as other (all other complaints listed in footnote a of Table 3). When oxygen was provided, the odds of MNA were higher (aOR ¼ 1.38) compared to when oxygen was not provided. The type of ambulance dispatched was also associated with MNA. Using BLS as a reference, the odds of MNA were higher for ALS (Advanced Life Support; refers to the type of care that extends past BLS to further assist in the injured or ill patient that includes "invasive" procedures 28 ). Level 1 (aOR ¼ 2.21), ALS Level 1, Emergency (aOR ¼ 2.12), and ALS Level 2 (aOR ¼ 3.29) service levels. Individuals who were treated and transported by EMS providers had the highest odds of MNA (aOR ¼ 1.62). Using home as the reference, the odds of MNA were lower when individuals were transported to the morgue/ mortuary (aOR ¼ 0.77) or any medical facilities (aOR ¼ 0.81). Lastly, after adjusting for demographic characteristics, clinical factors, and operational factors, the odds of an event requiring MNA significantly increased by 1.11 times on average each year (Table 3).

Discussion
This study found a substantial 54.3% relative increase in the percentage of opioid overdose events requiring MNA from 2015 to 2020 (from 18.4% in 2015 to 28.4% in 2020). This continues the trend observed from prior studies 19,24 and parallels the dramatic increase in rates of illicitly manufactured fentanyl or its analogues and the escalation of opioid overdoses during the COVID-19 pandemic. Given the significant and continuing rise in the need for MNA, this suggests that higher formulations of naloxone may fulfill an unmet need in addressing the opioid overdose crisis. The need for higher formulations of naloxone is further supported by the emerging body of evidence indicating that the use of naloxone often requires two or more doses to successfully reverse the effects of IMF overdose 29,30 due to its rapid toxic effects. 31,32 This study is strengthened by its use of a national database that covers a large sample size of 946,997 EMS events where naloxone was administered, from a broad geographic and demographic setting, and the longitudinal nature of the data allowing for assessment of naloxone administration trends over time and factors associated with MNA.
Stratification of the MNA results by US Census Regions revealed an increase in the percentage (as well as the odds) of events requiring MNA over the 5-year analysis timeframe within each region. However, there were some regional differences that corresponded with the differential trends in rates of opioid-related deaths due to synthetic opioids reported by the CDC Wide-Ranging Online Data for Epidemiology (WONDER) database. 7,33 The South exhibited a large increase in the rate of MNA from 2015 to 2018 (72.8%) which corresponded to the significantly larger (compared to the Northeast) increase in the rate of opioid-related deaths due to synthetic opioids by 126.3% from 2015 to 2018. Similarly, the slow but consistent rise in the rate of MNA in the West (17.4%) is likely due to the growing impact that synthetic opioids had in the West where a 188% increase in the rate of opioid-related deaths due to synthetic opioids was observed from 2015 to 2018. The smallest relative increase in the rate of MNA from 2015 to 2018 (13.9%) in the Northeast may be due to the early implementation of, and continued focus on opioid overdose prevention efforts given that this region exhibited the largest relative percent increase of opioid-related deaths due to synthetic opioids prior to 2015 (585% in the Northeast compared to 329%, 556% and 229% in the South, Midwest, and West, respectively from 2013 to 2015). Interestingly, although differential regional trends in the rate of MNA were exhibited, all regions have steadily increased to approximately 30% in 2020. This indicates that a significant percentage of the population requiring naloxone (approximately three in ten individuals) will need multiple administrations. The rate of MNA observed in 2020 may be an important starting point for EMS groups to consider for budgetary purposes. However, this assumes that the current mix of opioids is stabilizing. Unfortunately, illicit manufacturers are constantly experimenting with new compounds, thus a shift in the mix of opioids toward increased use of illicitly manufactured opioids may lead to a potentially higher rate of MNA in the future. Ultimately, given the growing number of individuals requiring multiple administrations, appropriate planning and distribution of higher dosage naloxone formulations may be needed to achieve timely and effective reversals of potentially lifethreatening opioid overdoses.
Compared to the analysis of 2012-2015 NEMSIS data conducted by Faul et al., the distribution of patient demographics, clinical factors, and operational factors were similar across the two studies. Specifically, those between 30-49 years of age still comprised the largest proportion in both studies, the proportion of males were also higher in both studies, approximately 50% of the events also occurred in the south in both studies, a significantly large (>80%) proportion of events still occurred in urban areas in both studies, oxygen was also provided in slightly less than half of the events in both studies, and more than 85% of the events still reported that individuals were treated & transported by EMS in both studies.
Given that there were very few events (<1%) that reported lay person administration of naloxone prior to EMS arrival in this study along with consistently higher rates of MNA by laypersons (from 43% to 83%) reported in prior studies, 29,30,34 the rate of MNA reported in this analysis is likely an underestimate. Plausible reasons for the paucity of events reporting lay person administration of naloxone include failure to keep accurate records, as well as failure of laypersons to report administering naloxone due to stigma-related challenges, including fear of shame, 35 fear  of law enforcement and related consequences of admitting that they are carrying naloxone, 36 and the general stigma 37 associated with opioid use. It is noted that most states in the US have passed Good Samaritan Laws (GSLs) which provide varying levels of legal immunity to laypersons who call for emergency assistance for overdose events. However, studies have shown that many persons who use drugs are not aware of the GSL and remain fearful of other consequences of calling for emergency assistance beyond a possession charge, including parole violation, child welfare factors, fear of losing housing, among others. [38][39][40] Additionally, regional/state differences in the adoption of Naloxone Accessibility Laws (NALs) and Opioid Education and Naloxone Distribution (OENDs) may also be a factor in the lower rates of MNA in this analysis compared to prior studies. Studies have shown that increased layperson possession and administration of naloxone has resulted in reduced opioid-related deaths. 15,41,42 Despite demonstrated improvements in reducing opioid-related deaths, a number of states have yet to adopt these initiatives, thus bystander access to naloxone may be limited. Although not reported in this study due to the small sample size, the rate of MNA by laypersons should also be explored to gain a broader understanding of the trend in naloxone administration. While prior studies 29,30,34 have reported consistently higher rates of MNA by laypersons (from 43% to 83%) compared to EMS, these studies had relatively small sample sizes and were conducted within limited geographic regions, therefore the results may not be generalizable.
The need for MNA, especially for bystanders, may increase as the influx of IMF and other highly potent synthetic opioids continue to escalate in the United States. These bystanders are likely to require more naloxone than EMS to reverse the overdose as they cannot provide supplemental oxygen or other interventions to manage the airway and provide ventilation. Given the critical role that bystanders may play during opioid overdose events, emphasis should be placed on understanding what bystanders are doing during these events in order to guide the refinement of initiatives including overdose education and take-home naloxone kits.
Current guidelines (e.g., World Health Organization, Centers for Disease Control and Prevention, Substance Abuse and Mental Health Services Administration, and American Academy of Emergency Medicine) regarding the optimal approach to treating opioid overdose encounters [43][44][45][46] do not provide recommendations that are empirically validated and were mostly developed before the surge in IMF and polysubstance use. Further research to characterize the different dosage, concentration, and route of administration of naloxone administered in various opioid overdose events can offer insights on developing an optimal protocol for naloxone administration.

Limitations
Limitations to this study include its retrospective observational evaluation of event information (submitted by EMS agencies in participating states) through convenience sampling thus the generalizability of the results may not extend to nonparticipating jurisdictions. In particular, a large number of states were not participating during 2017, likely due to issues in migration of data from version 2 to version 3, therefore the 2017 parameter estimates should be interpreted with the context that there were significant number of nonparticipating jurisdictions during 2017.
Furthermore, EMS protocol differences between and within states exist and may also affect the results. For example, some states only require the individual who overdosed to be revived only to the point where they are breathing, rather than fully conscious, which may affect the amount of naloxone that is administered to that person. Additionally, EMS protocol differences in how data was reported may have led to potential bias in the results due to missing data. For example, if one state allows for incomplete data on individuals when reporting treatment information, it may falsely appear to be utilizing fewer treatments than a state that diligently requires all treatment information to be recorded.
The NEMSIS Public Release Research data does not include clear & specific information regarding whether an event that required a naloxone administration was due to an opioid overdose event, the mode of naloxone administration (i.e., IV, IM, IN, etc.), or state identifiers, thus determination of whether these factors were specifically associated with changes in EMS naloxone administrations was not possible.
Substantial barriers exist to calling 911, particularly for people who use illicit opioids, due to the fear that they may be arrested for involvement with illegal drugs. As such, these individuals are most likely to be underrepresented relative to patients accidentally overdosing on prescription opioids. Given that the influx of illicitly manufactured fentanyl and other high-potency synthetic opioids are most likely driving the increase in MNA, the MNA rates reported in this study are likely to be underestimated. Accordingly, the need for higher doses of naloxone is also underestimated in this study.

Conclusion
The substantial 54.3% relative increase in the rate of MNA from 2015 to 2020 continues the upward trend observed in prior studies and reflects the evolving opioid crisis, particularly the dramatic increase in the rates of exposure to IMF or its analogues. With the accelerated increase in opioid-related deaths due to highly potent illicit synthetic opioids, in addition to expanding access to naloxone, higher naloxone formulations that can be rapidly and readily administered may be beneficial to combat the evolving opioid epidemic. Further research assessing differences in dosage, concentration, and routes of administration of naloxone administered in various opioid overdose events may offer insights on refinement of guidelines for optimal naloxone administration.

Authors' contributions
A. Raja Banerjee, Randa Abdelal, Neyla Darwaza, Chong Kim, Diane Ito and Josh Epstein were involved in the conception and design of this study, interpretation of the results and drafting/revising the paper for intellectual content. Chong Kim performed the analyses. Suzanne Carlberg Racich and Christopher Cebollero were involved in the interpretation of the results and drafting/revising the paper for intellectual content. All authors were involved in the final approval of this manuscript to be published and agree to be accountable for all aspects of the work herein. Note: The analysis excluded 5,982 events due to reporting unknown gender, unknown region and/or events occurring from the Island Areas. a Includes abdominal pain, allergies, animal bite, assault, back pain, breathing problem, burns, CO Poisoning/Hazmat, cardiac arrest, chest pain, choking, convulsions/seizure, diabetic problem, drowning, electrocution, eye problem, fall victim, headache, heart problems, heat/cold exposure, hemorrhage/laceration, industrial accident/inaccessible incident/other entrapments (nonvehicle), pregnancy/childbirth, psychiatric problem, sick person, stab/gunshot wound, stroke/CVA, traffic accident, traumatic injury, unconscious/fainting, unknown problem man down, transfer/interfacility/palliative care, mass casualty incident. b Includes fixed wing (airplane), paramedic intercept, specialty care transport, rotary wing (helicopter). c Includes assist (agency), assist (public), assist (unit), canceled, patient evaluated (no treatment/transport required), patient refused evaluation, standby, transport non-patient (organs, etc.). d Includes medical office/clinic, nursing home/assisted living facility, urgent care, and freestanding emergency department. e Other destination not specified in NEMSIS database.

Disclosure statement
Suzanne Carlberg-Racich and Christopher Cebollero have received consultancy fees from Hikma Community Health, a Subdivision of Hikma Specialty USA Inc. A. Raja Banerjee, Randa Abdelal, and Neyla Darwaza have received stock options exercisable for, and other stock awards of, ordinary shares of Hikma Pharmaceuticals in the course of their employment at Hikma Pharmaceuticals. Chong Kim, Diane Ito, and Josh Epstein are employees of Stratevi, a consulting firm that has received research funding from Hikma Community Health, a Subdivision of Hikma Specialty USA Inc., to conduct this study.

Funding
This review was funded by Hikma Community Health, a Subdivision of Hikma Specialty USA Inc. The funding organization was involved in the design of the study; interpretation of the data; preparation, review, and approval of the manuscript; and decision to submit the manuscript for publication.