Abstract
Background:
Missouri’s Overdose Field Report (ODFR) is a community-based reporting system which intends to capture overdoses which may not be otherwise recorded.
Objectives:
Describe the factors related to non-fatal overdoses reported to Missouri’s ODFR.
Methods:
This study used a descriptive epidemiological approach to examine the demographics and circumstances of overdoses reported to the ODFR. We used binary logistic regression to evaluate factors associated with survival and ordinal logistic regression to evaluate factors associated with number of doses used. Factors were chosen based on their relevance to overdose education and survival, and naloxone distribution.
Results:
Between 2018 and 2022, 12,225 overdoses (67% male; 78% White) were reported through the ODFR, with a 96% (n = 11,225) survival rate. Overdose survival (ps < .02) was associated with younger age (OR = .58), no opioid and stimulant co-involvement (OR = .61), intramuscular naloxone (OR = 2.11), and private location (OR = .48). An average of 1.6 doses of naloxone were administered. Additional doses were associated (ps < .02) with being older (OR = .45), female (OR = .90), nasal naloxone (versus intravenous) (OR = .65), and the belief fentanyl was present (OR = 1.49).
Conclusion:
Our reporting form provides a comprehensive picture of the events surrounding reported overdoses, including factors associated with survival, how much naloxone was used, and the effects of respondents believing fentanyl was involved. Missouri’s report can provide support for current naloxone dosing, contextualize refusing post-overdose transport, and can be used to improve overdose response by community and first responders.
Accurate and timely tracking of drug-related overdoses is essential for understanding and responding to the overdose epidemic. Existing infrastructure for overdose reporting in the United States is highly fragmented and focused on fatal overdoses. The State Unintentional Overdose Reporting System (SUDORS) incorporates data from death certificates, medical examiner or coroner reports, and postmortem toxicology data to record the number of fatal overdoses but has not been fully implemented across the country (1). Emergency services personnel respond to 7.3 non-fatal overdoses on average for each fatal overdose reported, which fatal overdose systems do not count (2). Non-fatal overdoses are recorded differently across states, localities, and individual emergency response agencies. Reporting systems only encounter a small proportion of non-fatal overdoses, depend on medical providers assumptions about a person experiencing overdose, and suffer from substantial reporting delays (3,4,5).
Supplementary reporting systems can address the under-reporting of non-fatal overdoses in existing surveillance systems. People who overdose may refuse post-overdose medical care due to concerns about healthcare-related costs, criminalization, and stigma. Despite Good Samaritan laws, community responders may be reluctant to contact emergency services due to fear of arrest (6,7,8). The number of community responders has increased in recent years with greater naloxone access, increasing the potential for unrecorded overdoses (9). By obtaining reports directly from community responders and people who experienced an overdose, we can “crowd-source” information about overdoses from the community. Community members, policy makers, care providers, and researchers can gain a better understanding of the hidden population of overdose survivors (3).
In the state of Missouri, the overdose crisis has resulted in thousands of lives lost. Missouri’s 2022 overdose death rate of 35.29 per 100,000 is among the highest west of the Mississippi River and exceeds the national average of 28.3 per 100,000 (10,11,12). Non-fatal overdoses are also common in Missouri, with overdoses in 2021 resulting in 177 emergency department visits per 100,000 and 134 inpatient hospitalizations per 100,000 (10). Prior to 2017, Missouri had no centralized system for EMS, public health officials, and members of the public to report overdose events and reversals. In 2017, the University of Missouri, St. Louis - Missouri Institute of Mental Health (UMSL-MIMH) created the anonymous Overdose Field Report (ODFR) as part of the Missouri Opioid-Heroin Overdose Prevention and Education (MO-HOPE) Project. Since its deployment, the ODFR has been used to monitor fatal and non-fatal overdoses across the state and inform naloxone distribution and harm reduction implementation efforts.
The ODFR has enabled data collection which had not been available through other means. Disseminated through overdose education and naloxone distribution (OEND) trainings, videos, fliers, and professional and community networks, the ODFR serves as a “crowd-sourced” survey mechanism to sample people who may witness an overdose. The ODFR collects information about the time, place, and possible drugs involved in an overdose, as well as information about both the person who overdosed and the person reporting the overdose. It has been adopted by community members, emergency responders, harm reduction organizations, and substance use service providers across the state. Emergency responders and people who use drugs are encouraged to complete ODFR reports in appreciation of continued access to free naloxone. Because the ODFR is available to professional and community responders alike, interaction with the medical or emergency response systems is not necessary for an overdose to be reported.
The aim of the present study is to describe overdose reversals in the state of Missouri between 2018 and 2022. We begin by describing the individuals who experienced overdoses, individuals who made a report to the ODFR, the location of reported overdoses, and the availability of transportation to a hospital or other location following an overdose. We also examined factors associated with two outcomes of interest: 1) survival of the person who experienced overdose, and 2) the number of naloxone doses per overdose administered. We focused on these two outcomes because survival is a key outcome for assessing overdose response, and the number of doses needed for a revival in the fentanyl era has been a source of recent debate (13). The factors chosen for these analyses include factors relevant to OEND training, including number of doses and type of naloxone, the demographics of the individual who overdosed, concern over the rise of fentanyl, and the concern that an overdose in a private residence may not be witnessed (14,15). These factors are associated with current and previous OEND and outreach initiatives to reduce overdose mortality within Missouri. These analyses are descriptive and exploratory, intended to describe the relationship between outcomes and predictors and are undertaken without specific hypotheses.
Methods
Target Population and Sample
This study is a retrospective analysis of overdose reports between 2018-01-01 and 2022-12-31 and has been approved by the University of Missouri—St. Louis institutional review board. Overdose reports submitted to the ODFR are completely self-report, with individuals who experienced, responded to, or witnessed an overdose completing the report afterwards. There are no required fields, meaning that respondents may skip any field they wish. Our target population is people who have experienced an overdose within our setting and period.
We used implicit inclusion sampling, meaning every entry was included when possible. We used listwise deletion for each regression analysis (see below) if responses were missing for the variable of interest. We describe our sample by age group, race, sex, and approximate location of event. In interpreting these data, it is important to note that the ODFR responses are self-report, not verified by medical or medical examiner data, and likely represent an undercount of overdoses which have occurred in areas where emergency responders and community members have not received OEND training, where naloxone distribution is limited, where internet access is poor, and among communities who fear drug-related criminalization.
The ODFR is disseminated primarily through overdose response training and naloxone distribution for emergency responders and community works, recovery community centers, and via mail-based naloxone distribution. St. Louis is both the epicenter of the overdose crisis in Missouri and the location for most of the overdose response trainings. As a result, ODFR data are clustered in the St. Louis area. Study data were collected and managed using REDCap electronic data capture tools hosted at the University of Missouri-St. Louis. REDCap (Research Electronic Data Capture) is a secure, web-based software platform designed to support data capture for research studies, providing 1) an intuitive interface for validated data capture; 2) audit trails for tracking data manipulation and export procedures; 3) automated export procedures for seamless data downloads to common statistical packages; and 4) procedures for data integration and interoperability with external sources (16,17).
Primary Outcomes of Interest
Survival
A single item asked whether the person who experienced the overdose survived. Respondents can report “Yes”, “No”, or “Unsure”. The structure of the ODFR does not provide additional detail about whether an individual may have died before or after naloxone administration.
Doses of Naloxone
Respondents were asked if naloxone was administered. If they reported it was, they were prompted to indicate what route (intramuscular, intravenous, nasal spray, or none), and the number of doses given (0 doses when “none” is selected). Doses of intramuscular and intravenous naloxone are assumed to be 0.4 mg, and doses of nasal naloxone are 4 mg, which are standard dosages. These doses are not indicated on the ODFR.
Other Characteristics of Interest
Respondents Reporting Overdoses
Respondents were asked to report their relationship to the person whose overdose has been reported. Available options were “Emergency Responder”, “Parent”, “Partner/Spouse”, “Other Family Member”, “Friend”, “Clinician/Provider”, “Stranger”, “Self”, and “Other”. When respondents select “Other”, they were prompted to provide additional information. We collapsed “Parent”, “Partner/Spouse”, and “Other Family Member” into a single “Family” category due to small group sizes; and “Clinician/Provider” and free text entries which indicated respondents were peer recovery coaches, counselors, and case workers into a single “Health Professional” category due to small groups sizes. “Other” responses were manually re-coded based on additional text provided, which included responses of “Neighbor”, “Roommate”, and entries which indicated “Other” and provided no additional information. Respondents who chose options other than “Self” were assumed to not be the person who experienced the overdose being reported. No other information about respondents is collected through the ODFR.
Location of Overdoses
Respondents entered the county in which the overdose occurred. Every county in Missouri and St. Louis City were available options. We recoded these based on their inclusion in the St. Louis Metro area (Crawford, Franklin, Jefferson, Lincoln, St. Charles, Warren, and St. Louis Counties and St. Louis City) compared to the rest of the state.
Hospital Transportation
When respondents reported a person survived, they were asked if the person received transport. Respondents are prompted to report if the person received transport to a hospital, an outpatient substance use treatment center, a residence, or “elsewhere”, if they declined transportation, it was unavailable, or if the respondent is unsure.
Covariates of Interest
We collected demographic information about the person who overdosed to the best knowledge of the party that completed the ODFR (age group, sex, race/ethnicity), the drugs suspected to be involved in overdose, and the scene where the overdose occurred. Response options for drugs involved included “heroin”, “prescription painkiller”, “fentanyl”, “methamphetamine”, “benzos”, “alcohol”, “other”, and “unsure”, which were recoded to “Opioids Alone”, “Opioids & Non-Stimulants Combined”, “Opioids & Stimulants Combined”, and “Other Drugs(s)”. Response options for scene of the overdose included “home/residence”, “treatment facility”, “public place”, and “other”. “Home/residence” was recoded to “Private”, and “treatment facility” was combined with “public place” due to few responses.
Analysis
We used descriptive statistics and counts to describe the sample and the responses to the ODFR on the outcome variables of interest. We evaluated predictors of survival by conducting a logistic regression in which we regressed survival (“yes” or “no”) on demographics, naloxone type and number of doses, type of drugs involved, and scene of overdose. We examined factors associated with the number of doses of naloxone administered for an overdose by using an ordinal logistic regression in which we regress the reported number of doses on naloxone type, drugs involved, demographics, and year of the overdose. For both analyses, missing data was treated using listwise deletion. All analyses were conducted in R, using the gt, tidyverse, and MASS packages (18,19,20,21).
Results
Sample Description
Between 2018 and 2022, there were 12,225 overdose events reported to the ODFR (Table 1). Of this sample, 68% of overdose events were among people between the ages of 25 and 44 (n = 7,997), 67% were male (n = 8,161), 78% were White (n = 9,308), and 98% were non-Hispanic/Latine (n = 10,288). Most people who experienced an overdose were Missouri residents (97%, n = 9,735). Overdoses were most common in private settings (65%, n = 7,638), and approximately half of all events involved transport to a hospital (48%, n = 5,187). Respondents reported suspected fentanyl involvement 56% of the time (n = 6,954), and a single substance was thought to be involved 79% of the time (n = 9,743). Other drugs reported to be involved in overdoses included stimulants (5.5%, n = 609), benzodiazepines (0.6%, n = 68), and alcohol (0.4%, n = 46). Intranasal naloxone was the most used type (68%, n = 6,898) and 10% of respondents reported multiple types of naloxone were used. The number of reports increased between 2018 and 2019, but dropped each following year, from 2661 in 2020 to 1566 in 2021. This drop in the total number of reports coincides with the beginning of the COVID-19 pandemic. Over the same three-year period, emergency responders (fire, police, and emergency medical services) became the most common respondents, making up greater than 40% of the reports completed between 2020 and 2022.
Table 1.
Overdoses Reported to the ODFR by Year in Missouri, 2018–2022, N = 12,224
| Total, N = 12,2241 | 2018, N = 2,3911 | 2019, N = 3,9041 | 2020, N = 2,6611 | 2021, N = 1,7031 | 2022, N = 1,5661 | |
|---|---|---|---|---|---|---|
|
| ||||||
| Responder | ||||||
| Stranger | 1,126 (9.2%) | 190 (7.9%) | 406 (10.4%) | 259 (9.7%) | 150 (8.8%) | 121 (7.7%) |
| Self | 594 (4.9%) | 174 (7.3%) | 265 (6.8%) | 81 (3.0%) | 29 (1.7%) | 45 (2.9%) |
| Friend | 4,145 (33.9%) | 1,054 (44.1%) | 1,792 (45.9%) | 658 (24.7%) | 322 (18.9%) | 319 (20.4%) |
| Family | 814 (6.7%) | 231 (9.7%) | 382 (9.8%) | 108 (4.1%) | 30 (1.8%) | 63 (4.0%) |
| Other2 | 150 (1.2%) | 27 (1.1%) | 56 (1.4%) | 20 (0.8%) | 21 (1.2%) | 26 (1.7%) |
| Emergency Responder | 4,012 (32.8%) | 532 (22.3%) | 711 (18.2%) | 1,139 (42.8%) | 944 (55.4%) | 686 (43.8%) |
| Health Professional | 1,384 (11.3%) | 183 (7.7%) | 292 (7.5%) | 396 (14.9%) | 207 (12.2%) | 306 (19.5%) |
| Age 3 | ||||||
| 25–44 | 7,956 (67.8%) | 1,613 (67.7%) | 2,714 (69.9%) | 1,580 (67.3%) | 1,048 (65.8%) | 1,001 (65.4%) |
| 45–64 | 1,840 (15.7%) | 365 (15.3%) | 540 (13.9%) | 396 (16.9%) | 277 (17.4%) | 262 (17.1%) |
| 65+ | 136 (1.2%) | 22 (0.9%) | 33 (0.9%) | 25 (1.1%) | 24 (1.5%) | 32 (2.1%) |
| 18–24 | 1,681 (14.3%) | 351 (14.7%) | 563 (14.5%) | 328 (14.0%) | 229 (14.4%) | 210 (13.7%) |
| Under 18 | 122 (1.0%) | 32 (1.3%) | 31 (0.8%) | 19 (0.8%) | 15 (0.9%) | 25 (1.6%) |
| Missing4 | 490 | 8 | 23 | 313 | 110 | 36 |
| Sex 3 | ||||||
| Female | 3,965 (32.8%) | 822 (34.7%) | 1,347 (34.7%) | 810 (31.0%) | 535 (31.8%) | 451 (28.9%) |
| Male | 8,110 (67.0%) | 1,543 (65.2%) | 2,517 (64.9%) | 1,803 (68.9%) | 1,144 (67.9%) | 1,103 (70.8%) |
| Other | 29 (0.2%) | 2 (0.1%) | 14 (0.4%) | 4 (0.2%) | 5 (0.3%) | 4 (0.3%) |
| Missing4 | 121 | 24 | 26 | 44 | 19 | 8 |
| Race 3 | ||||||
| Black | 2,512 (21.2%) | 388 (16.6%) | 635 (16.8%) | 722 (28.4%) | 384 (23.2%) | 383 (25.2%) |
| White | 9,239 (78.1%) | 1,938 (83.0%) | 3,107 (82.2%) | 1,807 (71.2%) | 1,262 (76.4%) | 1,125 (74.0%) |
| Other | 78 (0.7%) | 10 (0.4%) | 39 (1.0%) | 10 (0.4%) | 6 (0.4%) | 13 (0.9%) |
| Missing4 | 396 | 55 | 123 | 122 | 51 | 45 |
| Ethnicity 3 | ||||||
| Not Hispanic | 10,237 (97.9%) | 2,211 (98.0%) | 3,558 (97.8%) | 1,983 (98.6%) | 1,209 (96.9%) | 1,276 (97.8%) |
| Hispanic | 220 (2.1%) | 45 (2.0%) | 79 (2.2%) | 28 (1.4%) | 39 (3.1%) | 29 (2.2%) |
| Missing4 | 1,768 | 135 | 267 | 650 | 455 | 261 |
| Report from STL Metro Area | 8,716 (76.8%) | 2,036 (88.4%) | 3,175 (85.9%) | 1,652 (68.3%) | 841 (57.0%) | 1,012 (69.9%) |
| Missing4 | 883 | 89 | 206 | 243 | 227 | 118 |
| Transport by EMS | ||||||
| Refused | 3,308 (30.6%) | 775 (36.0%) | 1,154 (32.7%) | 543 (23.0%) | 401 (29.2%) | 435 (31.0%) |
| Hospital | 5,167 (47.8%) | 847 (39.4%) | 1,262 (35.7%) | 1,296 (55.0%) | 926 (67.3%) | 836 (59.6%) |
| Elsewhere | 806 (7.5%) | 259 (12.0%) | 414 (11.7%) | 58 (2.5%) | 18 (1.3%) | 57 (4.1%) |
| Treatment Center | 164 (1.5%) | 33 (1.5%) | 92 (2.6%) | 10 (0.4%) | 1 (0.1%) | 28 (2.0%) |
| Unavailable | 1,370 (12.7%) | 237 (11.0%) | 609 (17.2%) | 449 (19.1%) | 29 (2.1%) | 46 (3.3%) |
| Missing4 | 1,410 | 240 | 373 | 305 | 328 | 164 |
| Survival | 11,228 (95.6%) | 2,206 (94.9%) | 3,598 (95.2%) | 2,416 (96.4%) | 1,558 (96.2%) | 1,450 (95.7%) |
| Missing4 | 480 | 67 | 123 | 155 | 84 | 51 |
| Location | ||||||
| Private | 7,591 (65%) | 1,628 (68.6%) | 2,586 (66.8%) | 1,359 (58.1%) | 1,048 (63.8%) | 970 (63.1%) |
| Other5 | 1,070 (9.1%) | 190 (8.0%) | 340 (8.8%) | 255 (10.9%) | 137 (8.3%) | 148 (9.6%) |
| Public | 3,104 (26%) | 554 (23.4%) | 947 (24.5%) | 725 (31.0%) | 458 (27.9%) | 420 (27.3%) |
| Missing4 | 460 | 19 | 31 | 322 | 60 | 28 |
| Fentanyl | ||||||
| Not Present | 5,328 (43.6%) | 1,202 (50.3%) | 1,548 (39.7%) | 1,127 (42.4%) | 906 (53.2%) | 545 (34.8%) |
| Present | 6,897 (56.4%) | 1,189 (49.7%) | 2,356 (60.3%) | 1,534 (57.6%) | 797 (46.8%) | 1,021 (65.2%) |
| Drugs Involved | ||||||
| Opioids Alone | 9,394 (86.1%) | 1,937 (86.7%) | 3,169 (86.2%) | 1,993 (85.8%) | 1,106 (85.7%) | 1,189 (85.6%) |
| Opioids & Non-Stimulants | 723 (6.6%) | 200 (9.0%) | 271 (7.4%) | 158 (6.8%) | 51 (4.0%) | 43 (3.1%) |
| Stimulant & Opioids Combined | 522 (4.8%) | 69 (3.1%) | 177 (4.8%) | 102 (4.4%) | 69 (5.3%) | 105 (7.6%) |
| Other Drug(s) 6 | 274 (2.5%) | 28 (1.3%) | 59 (1.6%) | 70 (3.0%) | 65 (5.0%) | 52 (3.7%) |
| Missing4 | 1,312 | 157 | 228 | 338 | 412 | 177 |
| Naloxone Type | ||||||
| Nasal Naloxone | 6,856 (63.4%) | 1,053 (48.4%) | 2,065 (58.8%) | 1,623 (69.4%) | 1,162 (75.8%) | 953 (75.5%) |
| Intramuscular Naloxone | 1,758 (16.2%) | 617 (28.4%) | 708 (20.2%) | 254 (10.9%) | 100 (6.5%) | 79 (6.3%) |
| Multiple Types of Naloxone | 968 (8.9%) | 181 (9.1%) | 285 (8.8%) | 233 (11%) | 156 (11%) | 113 (9.6%) |
| Intravenous Naloxone | 472 (4.4%) | 145 (6.7%) | 167 (4.8%) | 89 (3.8%) | 38 (2.5%) | 33 (2.6%) |
| None | 767 (7.1%) | 180 (8.3%) | 286 (8.1%) | 139 (5.9%) | 77 (5.0%) | 85 (6.7%) |
| Missing4 | 1,404 | 215 | 393 | 323 | 170 | 303 |
| Number of Doses 3 | 1.60 (0.86) | 1.65 (0.90) | 1.66 (0.90) | 1.57 (0.81) | 1.51 (0.80) | 1.54 (0.84) |
n (%)
Demographic data for individual experiencing overdose, not respondent
Missing data are not included in percentage calculations
Mean (Standard Deviation)
Includes treatment facilities, street addresses, and responses of “other” without additional detail
Includes alcohol, stimulants alone, benzodiazepines, and responses of “other” without additional detail
Missing data are common in this type of data collection and 55% of records had missing data on at least one variable of interest. The total number of naloxone doses provided was missing most often, with 27% of cases lacking this information, followed by the type of naloxone used (17% of cases), overdose victims’ ethnicity (15% of cases), transportation after overdose (12% of cases), and the type(s) of drugs involved in the overdose (11% of cases). Sex, race, and age of the person who overdosed; overdose scene type; and whether the person survived were missing in less than 5% of reports.
Respondents
Respondents submitting the report included both professional and community responders. Friends were the most common respondents, representing 34% (n = 4145) of all overdose reports, followed by emergency responders (33%, n = 4012), and strangers (9%, n = 1138). People reported their own overdose 5% (n = 593) of the time. The remaining categories together totaled 480 reports (Partner/Spouse, 4%).
Location of Overdoses
Most responses originated from the St. Louis Metro area (71%, n = 8,615). In 2018 and 2019, more than 80% of responses indicated the overdose occurred in the St. Louis Metro area. As the ODFR became more widely disseminated, reports from the St. Louis Metro dropped to 68% (n = 1,652) in 2020, and 57% (n = 841) in 2021, but increased to 70% (n = 1,012) in 2022. Changes in where reports originated may be due to the spread of the ODFR via OEND trainings reaching more areas of the state, and appear to have been affected by the beginning of the COVID-19 pandemic.
Transportation
A hospital was the most common place where people were transported (n = 5187, 48%), followed by transportation refusal (n = 3335, 31%). Individuals who were transported to a hospital had a 99% survival rate, while all other forms of transport had a survival rate of at least 97%. In 2018 and 2019, transportation to the hospital and refusal occurred at similar rates, but beginning in 2020, transportation to the hospital became more common, exceeding 50% in each year, while refusing transportation became less common (2020: 23%; 2021: 29%; 2022: 31%). Being transported directly to a treatment center was reported in less than 3% of events over the entirety of the study period. Transportation was unavailable more often in 2019 and 2020, making up 17% (n = 613) and 20% (n = 467) of, respectively. Finally, Welch’s t-test showed that individuals who were transported to the hospital (M = 1.47, SD = 0.84) received fewer doses than those who were not transported to the hospital (M = 1.72, SD = 0.87), t(7666.3) = 13.4, p < .001.
Overdose Reversals and Survival
During the 5-year period, 93% (n = 11, 228) of reports indicated that the individual had survived, 4% (n = 517) did not survive, and 4% (n = 492) of reports did not include survival information. At no time during the study period did reports of survival drop below 90%.
A logistic regression found age, substances involved, scene, and naloxone type to be significantly associated with overdose survival (Table 2). The model included 7222 events, with 6987 survivals (97%). Events among people between 45–64-year-old (OR = 0.58, 95% CI = 0.44, 0.77; reference group: 25–44-year-old); which involved stimulants and opioids combined (OR = 0.61, 95% CI = 0.40, 0.95 reference group: Opioids Alone); which occurred in a private setting (OR = 0.48, 95% CI = 0.35, 0.65; reference group: Public Location), and events in which naloxone was not administered (OR = 0.16, 95% CI = 0.11, 0.24; reference group: nasal naloxone) had reduced chances of surviving the overdose. Compared to nasal naloxone, intramuscular naloxone was associated with an increase in odds of overdose survival (OR = 2.11, 95% CI = 1.40, 3.32).
Table 2.
Adjusted Odds Ratios for Logistic Regression Model Predicting Survival, N = 7,222
| Outcome: Survival (6987/7222) | OR1 | 95% CI1 | p-value |
|---|---|---|---|
|
| |||
| Race | |||
| White | — | — | |
| Black | 1.05 | 0.78, 1.42 | 0.77 |
| Other2 | 0.55 | 0.21, 1.91 | 0.30 |
| Age | |||
| 25–44 | — | — | |
| 45–64 | 0.58 | 0.44, 0.77 | <0.001 |
| 65+ | 0.66 | 0.26, 2.27 | 0.45 |
| 18–24 | 1.13 | 0.81, 1.60 | 0.50 |
| Under 18 | 1.29 | 0.50, 4.41 | 0.64 |
| Sex | |||
| Male | — | — | |
| Female | 0.97 | 0.77, 1.23 | 0.80 |
| Number of Doses | 0.91 | 0.76, 1.10 | 0.33 |
| Drugs Involved | |||
| Opioids Alone | — | — | |
| Opioids & Non-Stimulants | 0.79 | 0.54, 1.22 | 0.27 |
| Stimulant & Opioids Combined | 0.61 | 0.40, 0.95 | 0.02 |
| Other Drug(s)3 | 1.18 | 0.67, 2.20 | 0.59 |
| Fentanyl | |||
| Not Suspected | — | — | |
| Suspected | 0.98 | 0.77, 1.25 | 0.89 |
| Location | |||
| Public | — | — | |
| Private | 0.48 | 0.35, 0.65 | <0.001 |
| Other4 | 0.77 | 0.46, 1.32 | 0.33 |
| Naloxone Type | |||
| Nasal Naloxone | — | — | |
| Intramuscular Naloxone | 2.11 | 1.40, 3.32 | <0.001 |
| Multiple Types of Naloxone | 1.38 | 0.66, 3.56 | 0.44 |
| Intravenous Naloxone | 0.74 | 0.50, 1.12 | 0.14 |
| None | 0.16 | 0.11, 0.24 | <0.001 |
OR = Odds Ratio, CI = Confidence Interval
Includes Asian, Native American, and multiracial individuals.
Includes alcohol, stimulants alone, benzodiazepines, and responses of “other” without additional detail
Includes treatment facilities, street addresses, and responses of “other” without additional detail
Naloxone Doses
Between 2018 and 2022, the average number of naloxone doses reported was 1.60 per overdose (SD = 0.86), with a median of 2 (IQR = 1), and a minimum of 0 and a maximum of 6. Across all years, 3331 (27%) reports did not include information about the number of doses given. The average number of doses given by year fluctuated from a low of 1.51 (SD = 0.79) in 2021 and a high of 1.66 (SD = 0.90) in 2019. Of the cases which did not report the number of naloxone doses given, 1164 (35%) indicate the type of naloxone used. We used a logistic regression to investigate the relationship between naloxone type and missing information about the number of doses given, and found reports indicating intramuscular (OR = 1.42, 95% CI = 1.21, 1.66) and intravenous naloxone (OR = 6.21, 95% CI = 5.08, 7.59) were more likely to having missing information compared to nasal naloxone, while reports which indicated multiple types of naloxone were less likely to be missing the number of doses (OR = 0.61, 95% CI = 0.46, 0.80).
An ordinal logistic regression provided the best fit for the data compared to OLS, Poisson and negative binomial models based on AIC (ordinal = 14533.6, OLS = 15363.2, Poisson = 19922.7; NB = 19924.7) and BIC (ordinal = 14706.5, OLS = 15522.3, Poisson = 20074.8; NB = 20083.8), and the proportional odds assumption was met based on the Brant-Wald test (ps > .05), but results were similar across models (See Appendix). The ordinal regression (Table 3) included 7435 cases and showed that opioids in combination with other non-stimulant drugs were associated with greater doses of naloxone (OR = 1.27, 95% CI = 1.07, 1.50) compared to opioids alone. When fentanyl was believed to be involved in the overdose, more naloxone doses were administered (OR = 1.49, 95% CI = 1.35, 1.64). Administration of intravenous naloxone (OR = 0.65, 95% CI = −0.50, 0.85) was associated with fewer doses in comparison to intranasal naloxone. Overdose events in which emergency responders (OR = 0.44, 95% CI = 0.39, 0.50) and the individual who overdosed (OR = 0.81, 95% CI = 0.68, 0.95) completed the report were associated with fewer doses compared to when friends completed the report. When the individual was between the ages of 45 and 64, a greater number of doses was used (OR = 1.27, 95% CI = 1.11, 1.44) compared to individuals between the age of 25 and 44, while fewer doses were used when the individual was over the age of 65 (OR = 0.44, 95% CI = 0.25, 0.78). Finally, each consecutive year is associated with a small decrease in the number of doses of naloxone (0.93, 95% CI = 0.89, 0.97).
Table 3.
Adjusted Odds Ratios for Ordinal Regression Model Predicting Number of Naloxone Doses, N = 7,435
| Outcome: Number of Naloxone Doses | OR1 | 95% CI1 | p-value |
|---|---|---|---|
|
| |||
| Race | |||
| White | — | — | |
| Black | 0.95 | 0.85, 1.07 | 0.44 |
| Other2 | 1.66 | 0.89, 3.09 | 0.11 |
| Age | |||
| 25–44 | — | — | |
| 45–64 | 1.27 | 1.11, 1.44 | <0.001 |
| 65+ | 0.45 | 0.25, 0.77 | 0.004 |
| 18–24 | 0.89 | 0.78, 1.01 | 0.08 |
| Under 18 | 1.31 | 0.79, 2.16 | 0.29 |
| Sex | |||
| Male | — | — | |
| Female | 0.90 | 0.81, 0.98 | 0.02 |
| Drugs Involved | |||
| Opioids Alone | — | — | |
| Opioids & Non-Stimulants | 1.27 | 1.07, 1.50 | 0.007 |
| Stimulant & Opioids Combined | 1.07 | 0.86, 1.31 | 0.55 |
| Other Drug(s)3 | 0.71 | 0.50, 1.01 | 0.06 |
| Fentanyl | |||
| Not Suspected | — | — | |
| Suspected | 1.49 | 1.35, 1.64 | <0.001 |
| Naloxone Type | |||
| Nasal Naloxone | — | — | |
| Intramuscular Naloxone | 1.12 | 0.99, 1.28 | 0.07 |
| Intravenous Naloxone | 0.65 | 0.50, 0.85 | 0.002 |
| Multiple Types of Naloxone | 10.0 | 8.53, 11.8 | <0.001 |
| ODFR Respondent | |||
| Stranger | — | — | |
| Self | 0.81 | 0.68, 0.95 | 0.01 |
| Friend | 1.13 | 0.92, 1.39 | 0.24 |
| Family | 1.09 | 0.91, 1.30 | 0.35 |
| Other4 | 0.67 | 0.41, 1.08 | 0.10 |
| Emergency Responder | 0.44 | 0.39, 0.50 | <0.001 |
| Behavioral Health Professional | 0.86 | 0.65, 1.12 | 0.26 |
| Year | 0.93 | 0.89, 0.97 | <0.001 |
OR = Odds Ratio, CI = Confidence Interval
Includes Asian, Native American, and multiracial individuals.
Includes alcohol, stimulants alone, benzodiazepines, and responses of “other” without additional detail.
Includes neighbors and responses of “other” without additional detail.
Discussion
This study is a retrospective analysis of self-reported data about overdose responses and reversals in Missouri. Analyzing more than 12,000 overdose reports, we found most reports were submitted by friends of the person who had overdosed or by emergency responders, and most reported events occurred in the St. Louis Metro area. Individuals who overdosed were transported most often to a hospital following revival, with hospital transportation becoming more common as emergency responders became the primary group reporting overdoses. We found the combination of stimulants and opioids, the overdose occurring in a private location like a private residence, and no naloxone being used were associated with a lower probability of survival, while the use of intramuscular naloxone was associated with higher probability of survival. Finally, the number of naloxone doses used in a reported overdose decreased over time but remained between 1 and 2 doses on average.
During the 5-year study period, 12,225 reports were made to the ODFR. By comparison, the Missouri Department of Health and Human Services (DHSS) shows that there were 42,066 total inpatient visits for all drug overdoses and 14,071 opioid or stimulant involved overdose deaths during the same period, and more than 90,000 naloxone kits were dispensed in 2022 alone (10,22). In nearby Illinois, there were 92,184 overdoses reported between 2018 and 2022, with 78,580 (85%) of those being non-fatal (Illinois’ population is around double that of Missouri (23). These numbers suggest that Missouri’s ODFR underrepresents the number of overdoses which occurred over the study period, especially those which occurred outside of St. Louis.
The large number of non-fatal overdoses reported to the ODFR in which no subsequent medical attention was sought gives us insight into post-overdose care preferences. Professional first responders should be aware of the concerns people who use drugs may have that lead them to refuse transportation (24). By training first responders in harm reduction and post-overdose referral options, people who refuse transportation to a hospital may be more likely to receive the care they prefer at another type of service agency (25). People who experience overdose, as well as their friends and family, may be more comfortable contacting lay- and community-based responders like peers, individuals practicing street outreach, and harm reductionists (26,27).
The average number of naloxone doses used in an overdose event reported during this study remained at less than 2 doses per event. Studies conducted between 2000 and 2018 show a similar pattern, with both community and emergency responders using fewer than two doses in most successful overdose responses (28). While some events in our data indicated 3 to 6 doses were used, the data reported here suggest that high dose versions of naloxone may not be indicated in the overdose events described here.
Because most reported events did not receive hospital care, the experiences represented by ODFR reports may not be recorded through traditional systems. Our work shows the people at the heart of the overdose crisis can play an important role in data collection, which can complement existing surveillance systems by providing detail about non-fatal overdoses and individuals who overdose but do not encounter traditional systems (3). This timely information can inform harm reduction and treatment services and help local stakeholders to better respond to changing trends in overdoses. As new adulterants (e.g., Xylazine) enter the illicit drug supply, the capability to collect real-time data about overdoses can enable communities be aware of emerging trends, and accelerate their ability to respond (29).
Limitations and Future Research
As a descriptive study of an underreported phenomenon, this study is limited by biases associated with the ‘crowd-sourced’ nature of the survey method. The sample reported here may not fully represent the population of interest—overdose survivors in the state of Missouri—because sampling is limited to people who must be a) aware of the survey, b) have internet access via computer or smart phone, and c) have witnessed or survived an overdose. The number of reports began falling in 2020, which is associated with the beginning of the COVID-19 pandemic and the increase proportion of emergency responders completing reports. It is possible the pandemic resulted in fewer people present at overdoses resulted in fewer community responders reporting overdoses, but the current data do not allow further enquiry. There is also no method for follow-up reports which may provide information about post-overdose experiences. Some people also complete the ODFR an extended amount of time after the overdose occurred, resulting in inaccurate or biased reports. Without access to drug checking, it is also not possible for people to know which drugs are involved, exemplified by reports which do not include opioids at all or indicate the presence of fentanyl, despite the prevalence of synthetic opioids in the drug supply (30). The relatively small number of deaths due to overdose reported in this survey does not match the number reported by the Missouri Department of Health and Senior Services, which suggests these events are underreported (31). Finally, the dosage for each type of naloxone as stated in the methods is not on the ODFR form, and respondents may have a different understanding of what a single dose of naloxone is.
Future research should consider expansion of these methods to improve representation and the ability to monitor fatal and non-fatal overdoses. Data collection methods which are more accessible, including non-electronic forms, may provide an effective method of further reporting. Integrated reporting and harm reduction supply systems, like TxCOPE, can incentivize community and harm reduction organizations to distribute the survey form in communities most likely to experience or witness overdoses (3). By leveraging individual, community, state-, and federal-level data, it may be possible to better capture and use overdose and naloxone distribution data to understand geographic and temporal patterns of overdoses and overdose reversals. Finally, further investigation into who receives naloxone and how that naloxone is used and reported to the ODFR needs further investigation. This may require both new data collection methods and further investigation of the present data. With this information, harm reductionists, researchers, and community leaders can engage in targeted intervention to provide needed safety and drug checking supplies in areas experiencing the greatest need.
Conclusion
Current overdose reporting systems are reliant on traditional medical systems, and non-fatal overdoses. We demonstrate the utility of an informal, anonymous, publicly-accessible brief survey—the ODFR—to collect information about overdose events which is otherwise unavailable. Namely, through the Missouri ODFR, it appears that the number of doses of naloxone required to reverse overdoses has remained steady at 1–2 doses despite the proliferation of fentanyl in the drug supply, friends and emergency responders were the most likely groups to complete the overdose report, and more than half of overdose survivors decline hospital transport following the life-threatening overdose events. Missouri’s overdose reporting tool and the findings presented here can be used as a foundation for academic, government, and community organizations to adopt and refine with the overall goal of using data from overdose events to improve and customize overdose prevention efforts in their region.
Acknowledgments
The authors would like to thank Sarah Phillips, Katie Brown, and the entire Addiction Science team at the Missouri Institute of Mental Health at the University of Missouri—St. Louis.
Funding
Dr. Budesa, Mr. Vance, Mr. Smith, Mrs. Green, and Dr. Winograd are funded by the Substance Abuse and Mental Health Services Administration through the Prescription Drug Overdose (PDO) Grant (1H79SP022118) and the State Opioid Response grant (1H79TI085748–01), awarded to the Missouri Department of Mental Health, and while the work described in this article has been funded wholly or in part by the Missouri Department of Mental Health on behalf of the University of Missouri, St. Louis, it does not necessarily reflect the views of either agency. Dr. Banks was supported was supported by the National Institute on Drug Abuse (K08DA058080). Dr. Carpenter was supported was supported by the National Institute on Alcoholism and Alcohol Abuse (K23AA029729). Dr. Marshall, Dr. Schackman, and Dr. Zang were supported by the National Institute on Drug Abuse (U01DA047408).
Appendix
This appendix provides alternative model specifications for predicting the number of naloxone doses based on the predictors included in the manuscript.
Table A.1.
Model fit indices for Models predicting
| Model | AIC | BIC | R21 |
|---|---|---|---|
|
| |||
| Ordinal Logistic Regression2 | 14,533.64 | 14,706.48 | 0.48 |
| OLS Regression | 15,363.24 | 15,522.26 | 0.20 |
| Poisson Regression | 19,922.66 | 20,074.76 | 0.22 |
| Negative Binomial Regression | 19,924.73 | 20,083.76 | 0.22 |
R2 calculated as Nagelkerke’s R2 for Ordinal Logistic, Poisson, and Negative Binomial models, unadjusted R2 for OLS regression.
Ordinal Logistic regression treats outcome variable as ordered factor, other models treat it as numeric.
Table A.2.
OLS Regression Model Predicting Number of Naloxone Doses, N = 7,435
| Outcome: Number of Naloxone Doses | B | 95% CI1 | p-value |
|---|---|---|---|
|
| |||
| Race | |||
| White | — | — | |
| Black | −0.02 | −0.06, 0.02 | 0.402 |
| Other2 | 0.18 | −0.03, 0.40 | 0.100 |
| Age | |||
| 25–44 | — | — | |
| 45–64 | 0.08 | 0.03, 0.12 | <0.001 |
| 65+ | −0.24 | −0.42, −0.06 | 0.008 |
| 18–24 | −0.03 | −0.07, 0.01 | 0.191 |
| Under 18 | 0.11 | −0.07, 0.28 | 0.236 |
| Sex | |||
| Male | — | — | |
| Female | −0.04 | −0.08, −0.01 | 0.011 |
| Drugs Involved | |||
| Opioids Alone | — | — | |
| Opioids & Non-Stimulants | 0.09 | 0.03, 0.15 | 0.002 |
| Stimulant & Opioids Combined | 0.02 | −0.05, 0.10 | 0.519 |
| Other Drug(s)3 | −0.09 | −0.20, 0.02 | 0.117 |
| Fentanyl | |||
| Not Suspected | — | — | |
| Suspected | 0.14 | 0.10, 0.17 | <0.001 |
| Naloxone Type | |||
| Nasal Naloxone | — | — | |
| Intramuscular Naloxone | 0.05 | 0.01, 0.10 | 0.020 |
| Intravenous Naloxone | −0.12 | −0.21, −0.03 | 0.009 |
| Multiple Types of Naloxone | 0.91 | 0.86, 0.96 | <0.001 |
| ODFR Respondent | |||
| Stranger | — | — | |
| Self | −0.08 | −0.14, −0.02 | 0.007 |
| Friend | 0.06 | −0.01, 0.13 | 0.104 |
| Family | 0.04 | −0.02, 0.10 | 0.220 |
| Other4 | −0.12 | −0.29, 0.05 | 0.152 |
| Emergency Responder | −0.26 | −0.30, −0.22 | <0.001 |
| Behavioral Health Professional | −0.04 | −0.14, 0.05 | 0.384 |
| Year | −0.03 | −0.04, −0.01 | <0.001 |
CI = Confidence Interval
Includes Asian, Native American, and multiracial individuals.
Includes alcohol, stimulants alone, benzodiazepines, and responses of “other” without additional detail.
Includes neighbors and responses of “other” without additional detail.
Table A.3.
Poisson Regression Model Predicting Number of Naloxone Doses, N = 7,435
| Outcome: Number of Naloxone Doses | IRR1 | 95% CI1 | p-value |
|---|---|---|---|
|
| |||
| Race | |||
| White | — | — | |
| Black | 0.99 | 0.95, 1.04 | 0.65 |
| Other2 | 1.10 | 0.87, 1.37 | 0.42 |
| Age | |||
| 25–44 | — | — | |
| 45–64 | 1.05 | 1.00, 1.10 | 0.07 |
| 65+ | 0.85 | 0.68, 1.06 | 0.16 |
| 18–24 | 0.98 | 0.93, 1.03 | 0.51 |
| Under 18 | 1.06 | 0.87, 1.28 | 0.54 |
| Sex | |||
| Male | — | — | |
| Female | 0.97 | 0.94, 1.01 | 0.17 |
| Drugs Involved | |||
| Opioids Alone | — | — | |
| Opioids & Non-Stimulants | 1.05 | 0.98, 1.12 | 0.15 |
| Stimulant & Opioids Combined | 1.01 | 0.93, 1.09 | 0.79 |
| Other Drug(s)3 | 0.94 | 0.81, 1.07 | 0.34 |
| Fentanyl | |||
| Not Suspected | — | — | |
| Suspected | 1.08 | 1.04, 1.13 | <0.001 |
| Naloxone Type | |||
| Nasal Naloxone | — | — | |
| Intramuscular Naloxone | 1.04 | 0.99, 1.09 | 0.16 |
| Intravenous Naloxone | 0.93 | 0.83, 1.03 | 0.16 |
| Multiple Types of Naloxone | 1.55 | 1.48, 1.63 | <0.001 |
| ODFR Respondent | |||
| Stranger | — | — | |
| Self | 0.96 | 0.90, 1.02 | 0.19 |
| Friend | 1.03 | 0.95, 1.11 | 0.45 |
| Family | 1.02 | 0.95, 1.09 | 0.56 |
| Other4 | 0.94 | 0.77, 1.13 | 0.49 |
| Emergency Responder | 0.86 | 0.82, 0.90 | <0.001 |
| Behavioral Health Professional | 0.98 | 0.88, 1.08 | 0.69 |
| Year | 0.98 | 0.97, 1.00 | 0.06 |
IRR = Incident Rate Ratio, CI = Confidence Interval
Includes Asian, Native American, and multiracial individuals.
Includes alcohol, stimulants alone, benzodiazepines, and responses of “other” without additional detail.
Includes neighbors and responses of “other” without additional detail.
Table A.4.
Negative Binomial Regression Model Predicting Number of Naloxone Doses, N = 7,435
| Outcome: Number of Naloxone Doses | IRR1 | 95% CI1 | p-value |
|---|---|---|---|
|
| |||
| Race | |||
| White | — | — | |
| Black | 0.99 | 0.95, 1.04 | 0.65 |
| Other2 | 1.10 | 0.87, 1.37 | 0.42 |
| Age | |||
| 25–44 | — | — | |
| 45–64 | 1.05 | 1.00, 1.10 | 0.07 |
| 65+ | 0.85 | 0.68, 1.06 | 0.16 |
| 18–24 | 0.98 | 0.93, 1.03 | 0.51 |
| Under 18 | 1.06 | 0.87, 1.28 | 0.54 |
| Sex | |||
| Male | — | — | |
| Female | 0.97 | 0.94, 1.01 | 0.17 |
| Drugs Involved | |||
| Opioids Alone | — | — | |
| Opioids & Non-Stimulants | 1.05 | 0.98, 1.12 | 0.15 |
| Stimulant & Opioids Combined | 1.01 | 0.93, 1.09 | 0.79 |
| Other Drug(s)3 | 0.94 | 0.81, 1.07 | 0.34 |
| Fentanyl | |||
| Not Suspected | — | — | |
| Suspected | 1.08 | 1.04, 1.13 | <0.001 |
| Naloxone Type | |||
| Nasal Naloxone | — | — | |
| Intramuscular Naloxone | 1.04 | 0.99, 1.09 | 0.16 |
| Intravenous Naloxone | 0.93 | 0.83, 1.03 | 0.16 |
| Multiple Types of Naloxone | 1.55 | 1.48, 1.63 | <0.001 |
| ODFR Respondent | |||
| Stranger | — | — | |
| Self | 0.96 | 0.90, 1.02 | 0.19 |
| Friend | 1.03 | 0.95, 1.11 | 0.45 |
| Family | 1.02 | 0.95, 1.09 | 0.56 |
| Other4 | 0.94 | 0.77, 1.13 | 0.49 |
| Emergency Responder | 0.86 | 0.82, 0.90 | <0.001 |
| Behavioral Health Professional | 0.98 | 0.88, 1.08 | 0.69 |
| Year | 0.98 | 0.97, 1.00 | 0.06 |
IRR = Incident Rate Ratio, CI = Confidence Interval
Includes Asian, Native American, and multiracial individuals.
Includes alcohol, stimulants alone, benzodiazepines, and responses of “other” without additional detail.
Includes neighbors and responses of “other” without additional detail.
Footnotes
Disclosures
The authors report no relevant disclosures.
References
- 1.Centers for Disease Control and Prevention. CDC’s State Unintentional Drug Overdose Reporting System (SUDORS) [Internet]. 2022. [cited 2023 Mar 28]. Available from: https://www.cdc.gov/drugoverdose/fatal/sudors.html. [Google Scholar]
- 2.AZDHS | Opioid Prevention [Internet]. Ariz. Dep. Health Serv. [cited 2023 Dec 5]. Available from: http://www.azdhs.gov/opioid/. [Google Scholar]
- 3.Claborn K, Cance JD, Kane H, Hairgrove S, Conway FN. “If It Didn’t Get Reported, It Didn’t Happen”: Current Nonfatal Overdose Reporting Practices among Nontraditional Reporters in Texas. Subst Use Misuse. 2023;0:1–4. doi: 10.1080/10826084.2023.2188433. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Latimore AD, Newman J, Beletsky L. Build It Better for Public Health: Improved Data Infrastructure Is Vital to Bending the Curve of the Overdose Crisis. Am J Public Health. 2022;112:S39–S41. doi: 10.2105/AJPH.2021.306697. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Fliss MD, Cox ME, Dorris SW, Austin AE. Timely Overdose Death Reporting Is Challenging but We Must Do Better. Am J Public Health. 2021;111:1194–1196. doi: 10.2105/AJPH.2021.306332. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Madras BK, Ahmad NJ, Wen J, Sharfstein JS. Improving Access to Evidence-Based Medical Treatment for Opioid Use Disorder: Strategies to Address Key Barriers within the Treatment System. NAM Perspect. 2020: 10.31478/202004b. doi: 10.31478/202004b. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Bielenberg J, Swisher G, Lembke A, Haug N. A systematic review of stigma interventions for providers who treat patients with substance use disorders. J Subst Abuse Treat [Internet]. 2021. [cited 2023 Feb 1];131. doi: 10.1016/j.jsat.2021.108486. [DOI] [PubMed] [Google Scholar]
- 8.Lim JK, Forman LS, Ruiz S, Xuan Z, Callis BP, Cranston K, Walley AY. Factors associated with help seeking by community responders trained in overdose prevention and naloxone administration in Massachusetts. Drug Alcohol Depend. 2019;204:107531. doi: 10.1016/j.drugalcdep.2019.06.033. [DOI] [PubMed] [Google Scholar]
- 9.Watson DP, Ray B, Robison L, Huynh P, Sightes E, Walker LS, Brucker K, Duwve J. Lay responder naloxone access and Good Samaritan law compliance: postcard survey results from 20 Indiana counties. Harm Reduct J. 2018;15:18. doi: 10.1186/s12954-018-0226-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Missouri Department of Health & Senior Services. Drug Overdose Dashboard [Internet]. [cited 2023 Mar 9]. Available from: https://health.mo.gov/data/opioids/. [Google Scholar]
- 11.Zoorob M Fentanyl shock: The changing geography of overdose in the United States. Int J Drug Policy. 2019;70:40–46. doi: 10.1016/j.drugpo.2019.04.010. [DOI] [PubMed] [Google Scholar]
- 12.Centers for Disease Control and Prevention. Understanding the Opioid Overdose Epidemic [Internet]. 2022. [cited 2023 Mar 27]. Available from: https://www.cdc.gov/opioids/basics/epidemic.html. [Google Scholar]
- 13.Carpenter J, Murray BP, Atti S, Moran TP, Yancey A, Morgan B. Naloxone Dosing After Opioid Overdose in the Era of Illicitly Manufactured Fentanyl. J Med Toxicol. 2020;16:41–48. doi: 10.1007/s13181-019-00735-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Larochelle MR, Slavova S, Root ED, Feaster DJ, Ward PJ, Selk SC, Knott C, Villani J, Samet JH. Disparities in Opioid Overdose Death Trends by Race/Ethnicity, 2018–2019, From the HEALing Communities Study. Am J Public Health. 2021;111:1851–1854. doi: 10.2105/AJPH.2021.306431. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Fadanelli M, Cloud DH, Ibragimov U, Ballard AM, Prood N, Young AM, Cooper HLF. People, places, and stigma: A qualitative study exploring the overdose risk environment in rural Kentucky. Int J Drug Policy. 2020;85:102588. doi: 10.1016/j.drugpo.2019.11.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42:377–381. doi: 10.1016/j.jbi.2008.08.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Harris PA, Taylor R, Minor BL, Elliott V, Fernandez M, O’Neal L, McLeod L, Delacqua G, Delacqua F, Kirby J, et al. The REDCap consortium: Building an international community of software platform partners. J Biomed Inform. 2019;95:103208. doi: 10.1016/j.jbi.2019.103208. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.R Core Team. R: A language and environment for statistical computing. [Internet]. The R Foundation; 2022. [cited 2022 Dec 2]. Available from: https://www.r-project.org/. [Google Scholar]
- 19.Iannone R, Cheng J, Schloerke B, Hughes E, Lauer A, Seo J, Software P, PBC. gt: Easily Create Presentation-Ready Display Tables [Internet]. 2023. [cited 2023 Dec 5]. Available from: https://cran.r-project.org/web/packages/gt/index.html. [Google Scholar]
- 20.Wickham H, Averick M, Bryan J, Chang W, McGowan LD, François R, Grolemund G, Hayes A, Henry L, Hester J, et al. Welcome to the Tidyverse. J Open Source Softw. 2019;4:1686. doi: 10.21105/joss.01686. [DOI] [Google Scholar]
- 21.Venables WN, Ripley BD. Modern Applied Statistics with S [Internet]. New York, NY: Springer; 2002. [cited 2023 Dec 5]. Available from: http://link.springer.com/10.1007/978-0-387-21706-2. [Google Scholar]
- 22.Vance K, Park B, Kondai R, Budesa Z, Winograd RP. Missouri Naloxone Distribution from 2017–2022: Evaluation and Implications of Applying a Naloxone Saturation Model. n.d.;In process. [Google Scholar]
- 23.Opioid Data Dashboard [Internet]. [cited 2023 Dec 20]. Available from: https://idph.illinois.gov/OpioidDataDashboard/. [Google Scholar]
- 24.Meyerson BE, Russell DM, Kichler M, Atkin T, Fox G, Coles HB. I don’t even want to go to the doctor when I get sick now: Healthcare experiences and discrimination reported by people who use drugs, Arizona 2019. Int J Drug Policy. 2021;93:103112. doi: 10.1016/j.drugpo.2021.103112. [DOI] [PubMed] [Google Scholar]
- 25.Siddiqui S, La Manna A, Connors E, Smith R, Vance K, Budesa Z, Woods C, Goulka J, Beletsky L, Winograd R. Responding and Referring: A Within-Person Evaluation of First Responders’ Intention to Refer to Post Overdose Services following SHIELD Training. Unpubl Data. 2023; [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Mehta A, Xavier JC, Palis H, Slaunwhite A, Jenneson S, Buxton JA. Change in Police Attendance at Overdose Events following Implementation of a Police Non-Notification Policy in British Columbia. Adv Public Health. 2022;2022:e8778430. doi: 10.1155/2022/8778430. [DOI] [Google Scholar]
- 27.Ray B, Hedden B, Carroll JJ, Del Pozo B, Wagner K, Kral A, O’Donnll D, Victor G, Huynh P. Establishing a Non-Police, Community-Based Crisis Response Team as a Primary Responder in Charlotte: Stakeholder Feedback and Development Report [Internet]. Promise Resource Network; 2021. [cited 2023 Jul 20]. Available from: https://www.promiseresourcenetwork.org/wp-content/uploads/2021_PRNxSAFE_Crisis_Intervention_Unit_Stakeholder_Report.pdf. [Google Scholar]
- 28.Rzasa Lynn R, Galinkin J. Naloxone dosage for opioid reversal: current evidence and clinical implications. Ther Adv Drug Saf. 2018;9:63–88. doi: 10.1177/2042098617744161. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Friedman J, Montero F, Bourgois P, Wahbi R, Dye D, Goodman-Meza D, Shover C. Xylazine spreads across the US: A growing component of the increasingly synthetic and polysubstance overdose crisis. Drug Alcohol Depend. 2022;233:109380. doi: 10.1016/j.drugalcdep.2022.109380. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Ciccarone D The rise of illicit fentanyls, stimulants and the fourth wave of the opioid overdose crisis. Curr Opin Psychiatry. 2021;34:344–350. doi: 10.1097/YCO.0000000000000717. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Missouri Department of Health & Senior Services. Quarterly data request DHSS [Unpublished Raw Data]. Missouri Department of Health & Senior Services; 2022. [Google Scholar]