Leveraging Body-Worn Camera Footage to Better Understand Opioid Overdoses and the Impact of Police-Administered Naloxone

Abstract

Objectives. To investigate what transpires at opioid overdoses where police administer naloxone and to identify the frequency with which concerns about police-administered naloxone are observed.

Methods. We reviewed body-worn camera (BWC) footage of all incidents where a Tempe, Arizona police officer administered naloxone or was present when the Tempe Fire Medical Rescue (TFMR) administered it, from February 3, 2020 to May 7, 2021 (n = 168). We devised a detailed coding instrument and employed univariate and bivariate analysis to examine the frequency of concerns regarding police-administered naloxone.

Results. Police arrived on scene before the TFMR in 73.7% of cases. In 88.6% of calls the individual was unconscious when police arrived, but 94.6% survived the overdose. The primary concerns about police-administered naloxone were rarely observed. There were no cases of improper naloxone administration or accidental opioid exposure to an officer. Aggression toward police from an overdose survivor rarely occurred (3.6%), and arrests of survivors (3.6%) and others on scene (1.2%) were infrequent.

Conclusions. BWC footage provides a unique window into opioid overdoses. In Tempe, the concerns over police-administered naloxone are overstated. If results are similar elsewhere, those concerns are barriers that must be removed. (Am J Public Health. 2022;112(9):1326–1332. https://doi.org/10.2105/AJPH.2022.306918)

The opioid overdose crisis emerged 30 years ago, but the last 5 years have been especially deadly. ¹ During the COVID-19 pandemic in 2020, the United States experienced more than 93 000 overdose deaths, nearly 70 000 of which involved opioids. ² The number of overdose deaths increased again from May 2020 to April 2021 to more than 100 000, with synthetic opioids causing 64% of them. ³

Many jurisdictions have publicly distributed naloxone as a response to the opioid overdose crisis. Naloxone reverses an overdose by binding to receptors in the brain, thereby reducing the chances of brain damage and restoring “normal breathing.” ^4(p1202) Naloxone has been available for 25 years; by 2015, more than 27 000 lives had been saved with naloxone. ⁵

The police often respond to opioid overdoses, given their availability, rapid response, and duty to protect life. ⁶ In 2014, the US Department of Justice ⁷ created the Law Enforcement Naloxone Toolkit to support the adoption of naloxone programs by police. Several studies show that police can safely administer naloxone. ^{8, 9} However, by 2019, only 2500 of the nearly 18 000 US law enforcement agencies had deployed naloxone to their officers. ¹⁰

Diffusion of police-administered naloxone has been slow, in part because of cost ¹¹ and recent efforts to divert overdose calls away from police response. ¹² Police have also expressed concerns about naloxone, including negative attitudes about people who use drugs, ^{13, 14} worries about increased civil or criminal liability and administering naloxone improperly, ^{15, 16} fatigue from responding to overdoses, ⁹ and fears of accidental exposure to opioids and aggression from recovering individuals. ^{17, 18}

Some public health and harm reduction experts and those who use drugs have noted that police involvement will criminalize overdoses through the arrest of survivors, and several studies support this concern. ^{19, 20} One study showed that police response was associated with an increased rate of nonfatal overdoses. ²¹ These concerns are especially acute in states with no Good Samaritan Laws that provide immunity from arrest for overdose survivors and others on scene. ²² Drug-induced homicide laws raise similar concerns and can lead to reluctance to dial 911. ²³ Prior research has not sufficiently investigated concerns over police-administered naloxone.

METHODS

We addressed these questions through an examination of body-worn camera (BWC) footage of 168 cases in which Tempe, Arizona police officers administered intranasal naloxone or witnessed Tempe Fire Medical Rescue (TFMR) personnel deliver the drug, from February 3, 2020 to May 7, 2021 (n = 168). BWC footage provides a unique window into what occurs during opioid overdoses. It also allows examination of the frequency with which concerns about police-administered naloxone are observed.

The Tempe Police Department (TPD) employs 350 sworn officers. Officers began carrying intranasal naloxone in January 2020 as part of a project that funded the purchase of naloxone for both police and community members. ²⁴ Officers have carried BWCs since May 2016, and the department requires officers to record all potential drug overdoses.

The TPD granted us access to BWC footage through the department’s data storage system. One of the authors (C. O.) is a crime analyst with the TPD and has permanent secure access. Two other authors (S. W., M. D. W.) became TPD volunteers, which also guaranteed access. TPD officers had a 100% activation rate for all cases during the study period. In 1 case, an officer muted the BWC audio, which prevented coding. The case has been excluded.

We devised a coding instrument to capture 157 variables describing characteristics of the overdose, naloxone administration, officer and citizen behavior, and outcomes. We based the instrument on a tool used to code BWC footage on a previous project evaluating de-escalation training. Two coders, with significant training and experience in coding BWC footage, independently coded all cases. We used Stata version 15.1 (StataCorp LLC, College Station, TX) to compare every variable coded in the 2 independent data sets. We found coding discrepancies in 10% of the approximately 26 000 variables coded across the 168 overdoses; we resolved the discrepancies prior to analysis. For some variables, the reported sample size was less than 168 because we were not able to capture information from the BWC.

The TPD routinely collects identifying information for everyone at an overdose call to run names through all available databases. This information-gathering process may lead to the identification of outstanding warrants. Because BWC footage only captures what transpires at the scene in the camera’s view, we ran each case through the TPD records management system to determine whether anyone on scene was subsequently ticketed or arrested.

We used univariate and bivariate analyses to address our research questions. First, we examined what occurred when police administered naloxone (e.g., aspects of the scene, who was present, characteristics of naloxone administration, survival). Second, we tested the primary concerns regarding police-administered naloxone:

• 1.Was there any indication that officers hesitated to administer naloxone (officer did not immediately administer naloxone after checking for signs of overdose)?
• 2.Did any officers experience accidental exposure to opioids?
• 3.How often did officers improperly administer naloxone (officer did not follow training protocols)?
• 4.How often did overdose survivors show aggression toward officers after recovery (physically combative or resistant behavior requiring a force action by the officer)?
• 5.How frequently did officers express negative attitudes or treatment toward the overdose survivor?
• 6.How frequently were overdose survivors arrested?
• 7.How frequently were others on scene arrested?
• 8.Were any officers disciplined, sued, or criminally charged for administering naloxone? (Tempe Police Department Narcan program coordinator, written communication, March 23, 2022)

RESULTS

On the basis of BWC footage, we determined that in 97.6% of encounters, the officer became involved because of a dispatched call for service (Table 1). On average, 2 other people were at the scene, including family members (24.6%), friends (44.9%), and bystanders (30.1%). We determined the relationships among people from BWC-recorded communication between the officer and those on scene. The person experiencing the overdose was alone in just 11.8% of the incidents. The average response time was 5:01 minutes, and the officer arrived on scene before the TFMR in 73.7% of incidents. In cases where the TPD arrived first, we were not able to consistently capture the TFMR arrival time. Police officers administered naloxone in 74.1% of encounters (the TFMR conducted the rest). Police administered multiple doses in 51.8% of cases. The mean time between first and second dose was 54 seconds. This is well short of the recommended wait time (3–4 minutes). Police performed CPR (cardiopulmonary resuscitation) in 33.5% of cases.

TABLE 1—

Characteristics of Naloxone Administration Incidents: Tempe, AZ, February 3, 2020–May 7, 2021

Variable	No. (%)
How officer became aware of call for service
Dispatched call for service	164 (97.6)
Other (e.g., citizen flag down)	4 (2.4)
Was person experiencing overdose alone?
Yes	20 (11.8)
No	149 (88.0)
Family members on scene?
Yes	41 (24.6)
No	126 (75.4)
Friends on scene?
Yes	75 (44.9)
No	92 (55.1)
Bystanders or others on scene?
Yes	50 (30.1)
No	116 (69.9)
Fire department on scene before TPD?
Yes	44 (26.4)
No	123 (73.7)
No. of TPD naloxone administrations on scenea
0	43 (25.9)
1	37 (22.3)
≥ 2	86 (51.8)
Officer performed CPR
Yes	52 (33.5)
No	103 (66.5)

Note. CPR = cardiopulmonary resuscitation; TPD = Tempe Police Department.

^a Mean number of administrations = 1.43 (SD = 1.14). When police administered 2 doses, the mean time between doses was 54 seconds.

Table 2 shows that three quarters of overdose victims (76.6%) were men, and 74.4% were aged 18 to 39 years. Fifty-five percent were White, 21.0% were Black, and 16.2% were Latinx. Most individuals (88.6%) were not conscious when the police arrived on scene. Nearly all individuals experiencing an overdose survived (94.6%), and 84.4% were transported to a hospital. As shown in Table 3, there is little evidence to support the primary concerns about police-administered naloxone.

TABLE 2—

Characteristics of Persons Experiencing Drug Overdose During Naloxone Administration Incidents: Tempe, AZ, February 3, 2020–May 7, 2021

Variable	No. (%)
Gender
Male	128 (76.6)
Female	39 (23.4)
Age, y
≤ 17	8 (4.8)
18–29	70 (41.7)
30–39	55 (32.7)
40–74	34 (20.2)
≥ 75	1 (0.6)
Race/ethnicity
White	92 (55.1)
Black	35 (21.0)
Latinx	27 (16.2)
Other	13 (7.8)
Victim conscious upon PD arrival?
Yes	19 (11.4)
No	148 (88.6)
Verbal confirmation of opioid use?
Yes	109 (67.7)
No	52 (32.3)
Opioid type (reported by someone at scene)
Fentanyl	25 (14.9)
Heroin	12 (7.1)
Percocet, OxyContin, oxycodone, Vicodin, M30s, “blue” pills	35 (20.8)
Other	14 (8.3)
Combination of drugs	24 (14.3)
Unclear, not specified	58 (34.5)
Did the individual survive?
Yes	159 (94.6)
No	9 (5.4)
Was overdose victim transported to hospital?
Yes—taken by FD	135 (84.4)
No—refused	19 (11.9)
No—released at the scene	6 (3.8)

Notes. FD = fire department; PD = police department.

TABLE 3—

Concerns About Naloxone Administration: Tempe, AZ, February 3, 2020–May 7, 2021

Variable	No. (%)
Police concerns
Did the officer hesitate to administer naloxone?
Yes	1 (0.6)
No	120 (71.9)
Unclear	46 (27.5)
Did the officer improperly administer naloxone?
No	168 (100)
Was an officer accidently exposed to an opioid?
No	168 (100)
Did the overdose victim behave aggressively?
Yes	6 (3.6)
No	162 (96.4)
Was an officer disciplined, sued, or criminally charged?
No	168 (100)
Public health concerns
Was the overdose victim arrested?
Yes	6 (3.6)
Noa	162 (96.4)
Was the overdose victim ticketed or cited?
Yes	5 (3.0)
No	159 (97.0)
Was anyone else on scene arrested?
Yes	2 (1.2)
No	166 (98.8)
Was anyone else on scene ticketed or cited?
No	168 (100)
Did officer act impersonally (cold or indifferent)?
Yes	2 (1.2)
No	166 (98.8)
Did officer use condescending or patronizing tone?
Yes	2 (2.0)
No	100 (98)
Did officer yell or raise voice at overdose survivor?
No	168 (100)
Was officer reactive, angry, or abrasive?
Yes	1 (0.8)
No	128 (99.2)

^a In 8 cases, the person experiencing the overdose had a warrant, but the officer(s) on scene did not arrest (4.7%).

Police Concerns

Among the 168 incidents, no officers improperly administered naloxone. No officers were accidentally exposed to opioids, and just 1 officer (0.6%) hesitated to use naloxone. No officers were subsequently disciplined, sued, or criminally charged because of a naloxone administration. In only 3.6% of cases did individuals experiencing an overdose become aggressive after being revived.

Public Health Concerns

Police arrested 3.6% of the individuals experiencing an overdose (all were arrested because of an outstanding felony warrant). In an additional 8 cases, the person experiencing the overdose had a warrant, but the police did not make an arrest (4.7%). The person experiencing the overdose was ticketed or cited in 3.0% of cases. Others on scene were rarely arrested (1.2%) or ticketed or cited (0%). Police rarely used negative treatment or expressed negative attitudes toward the person, by acting impersonally (cold or indifferent; 1.2% of cases), using condescending or patronizing language (2.0%), yelling (0%), or being reactive, angry, or abrasive (0.8%).

DISCUSSION

BWC footage allows us to examine the frequency with which concerns regarding police-administered naloxone are observed. We found that in Tempe, Arizona, during the period of our study, these concerns were unfounded. No officer improperly administered naloxone, and in only 1 case, the officers hesitated to administer naloxone. No officers experienced an accidental exposure to opioids. Aggression toward police rarely occurred (3.6%), and police rarely acted impersonally (1.2%), used a condescending or patronizing tone (2.0%), yelled (0%), or displayed anger (0.8%). No officers experienced negative administrative, civil, or criminal consequences after delivering naloxone. Police arrested 6 survivors (3.6%) and 2 others on scene (1.2%).

There are context-specific explanations for these findings. In 2015, the Arizona State Legislature passed House Bill 2489, which permits police with proper training to administer naloxone. ²⁵ The Arizona Governor’s Office ²⁶ and the Arizona Peace Officer Standards and Training Board ²⁷ have promoted police-administered naloxone through funding, training, and access to naloxone. In 2018, the Arizona State Legislature passed the “911 Good Samaritan Law,” which provides protections against arrest (outstanding warrants are not included) for the person who overdosed and the person calling for medical assistance. ²⁸ The TPD’s training highlights state law and de-emphasizes arrest in overdose calls. ²⁹ In sum, training, legal mandates, and support in the police department and community provided the necessary ingredients for an effective police-administered naloxone program in Tempe. Police departments without these ingredients may not be well suited to carry naloxone.

Limitations

The current study suffers from several limitations. We studied 1 jurisdiction in the southwestern United States during a 15-month period. These results may not be generalizable to other jurisdictions, especially given the recent increase in overdoses among African Americans. ^{20, 30} The study also coincided with the COVID-19 pandemic, which may have influenced police response to opioid overdoses. We relied on BWC footage as our primary data source. We are unable to comment on overdoses where there was no call to 911 and no police response, or no use of naloxone. BWC footage captures real-time audio and video at opioid overdoses scenes, but the cameras have limitations, such as problems with dim lighting and camera angle. ³¹ We were not able to document anything that occurred outside the capture area of the BWC (e.g., a records check, arrival time of the TFMR), or that occurred after the encounter ended.

We gained access to the BWC data as a TPD employee and TPD-approved volunteers. Access to such data is often restricted and can be difficult to obtain. Last, some of the coding from BWC footage is inherently subjective. We attempted to minimize this issue through rigorous training of our coders and a double-blind coding scheme with cell-by-cell data comparison for reliability.

Public Health Implications

Although more research is needed, the evidence we present here offers support for police-administered naloxone as a potentially effective response to opioid overdoses. The results also highlight important guidelines for police departments to follow. The effort should be grounded in harm reduction, de-emphasize arrest, and provide training that prepares officers to diagnose an opioid overdose, administer naloxone to avoid acute precipitated withdrawal (3–4 minutes between doses), and engage with survivors. It should be coupled with efforts to distribute naloxone to first responders, people who use drugs, and their family and friends (bystanders were present in 88% of overdoses examined here). ³² Local and state governments should pass Good Samaritan Laws that protect from arrest overdose survivors and those who contact emergency services.

Law enforcement, public health experts, and researchers should continue to investigate opioid overdoses and the concerns surrounding police-administered naloxone. Recent studies highlight the pervasiveness of these concerns, the lack of evidence to support the claims, ³³ and the impact of a brief training on dispelling those “false beliefs.” ^17(p34) If the evidence is as strong in other jurisdictions as it is in Tempe, those concerns are barriers that must be removed.