Presenting Rhythm in Sudden Deaths Temporally Proximate to Discharge of TASER Conducted Electrical Weapons

Abstract

Objectives:  Sudden deaths proximate to use of conducted electrical weapons (CEWs) have been attributed to cardiac electrical stimulation. The rhythm in death caused by rapid, cardiac electrical stimulation usually is ventricular fibrillation (VF); electrical stimulation has not been reported to cause asystole or pulseless electrical activity (PEA). The authors studied the presenting rhythms in sudden deaths temporally proximate to use of TASER CEWs to estimate the likelihood that these deaths could be caused by cardiac electrical stimulation.

Methods:  This was a retrospective review of CEW-associated, nontraumatic sudden deaths from 2001 to 2008. Emergency medical services (EMS), autopsy, and law enforcement reports were requested and analyzed. Subjects were included if they collapsed within 15 minutes of CEW discharge and the first cardiac arrest rhythm was reported.

Results:  Records for 200 cases were received. The presenting rhythm was reported for 56 of 118 subjects who collapsed within 15 minutes (47%). The rhythm was VF in four subjects (7%; 95% confidence interval [CI] = 3% to 17%) and bradycardia-asystole or PEA in 52 subjects (93%; 95% CI = 83% to 97%). None of the eight subjects who collapsed during electrocardiogram (ECG) monitoring had VF. Only one subject (2%) collapsed immediately after CEW discharge. This was the only death typical of electrically induced VF (2%, 95% CI = 0% to 9%). An additional 4 subjects (7%) collapsed within 1 minute, and the remaining 51 subjects (91%) collapsed more than 1 minute later. The time from collapse to first recorded rhythm was 3 minutes or less in 35 subjects (62%) and 5 minutes or less in 43 subjects (77%).

Conclusions:  In sudden deaths proximate to CEW discharge, immediate collapse is unusual, and VF is an uncommon VF presenting rhythm. Within study limitations, including selection bias and the possibility that VF terminated before the presenting rhythm was recorded, these data do not support electrically induced VF as a common mechanism of these sudden deaths.

Nontraumatic, sudden deaths of civilians during interactions with law enforcement are a public health concern.¹ The TASER conducted electrical weapon (CEW; TASER International Inc., Scottsdale, AZ) is used increasingly by law enforcement officers because of its unique capability to incapacitate violent or resisting subjects. Deaths temporally related to use of CEWs have attracted attention from the media² and medical organizations.³

Conducted electrical weapons deliver rapid electrical stimulation at frequency of 15–20 Hz to incapacitate subjects by causing contraction of skeletal muscles,⁴ using either two tethered barbed-probe projectiles or two closely spaced electrodes on the cartridge of the hand-held unit (“drive-stun” mode). In all reported animal experiments and clinical cases of cardiac arrests caused by rapid, cardiac electrical stimulation from other sources, the rhythm has been ventricular fibrillation (VF).^5,6 Rapid cardiac stimulation has not been reported to cause asystole or pulseless electrical activity (PEA).^5,7 Thus, investigators have postulated^8–14 and media reports have implied that deaths proximate to use of CEWs are caused by electrically induced VF.

If CEWs cause death by cardiac stimulation, the initial postarrest rhythm should be VF, and subjects should collapse within 10 seconds after CEW discharge. However, if either collapse is delayed or the initial rhythm is asystole or PEA, then electrically induced VF is not the cause of death; thus, cardiac stimulation by the CEW cannot be implicated as the direct cause of death.

Data regarding the cardiac rhythms in deaths proximate to use of CEWs are limited. Our objective was to identify the first-recorded rhythms in these sudden deaths as an estimate of the initial, cardiac arrest rhythms, to estimate the likelihood that they could be caused by cardiac electrical stimulation.

Methods

Study Design

This was a retrospective survey of CEW-associated, nontraumatic sudden deaths identified through an Internet search. The study was performed under a letter of assent by the Cedars Sinai Medical Center (Los Angeles, CA) Institutional Review Board.

Study Setting and Population

We performed an Internet-based search for deaths temporally proximate to use of TASER CEWs via the Google search engine, using key words “TASER” AND “death.” Subsequently, we performed searches using other key words (electronic control device, conducted electrical weapon, TAZER, electric gun, stun gun) AND “death.” Subjects who died suddenly and unexpectedly after law enforcement officers discharged a CEW were included based on these prospective criteria: 1) the discharge was applied to the subject; 2) the subject became unresponsive (“collapsed”) within 15 minutes of the last discharge; 3) EMS, automatic external defibrillator (AED), or medical records included a cardiac rhythm diagnosis prior to treatment with drugs or defibrillation; and 4) death was not caused by physical trauma unrelated to the CEW discharge. We obtained an autopsy report for all subjects who met these four criteria.

Study Protocol

We made written requests for records of deaths between January 1, 2001, and February 1, 2008, from appropriate coroners, medical examiners, law enforcement agencies, and emergency medical services (EMS) under the Freedom of Information Act. If we received no response, we followed-up with a reminder letter and phone call. We specifically requested electrocardiograms (ECGs), recordings from AEDs, video or audio recordings, downloads of CEW data, and autopsies. We recorded demographic data from EMS and autopsy reports. Toxicologic data were recorded from autopsy reports whenever available. The CEW model and type of electrodes were recorded whenever they were reported.⁴

Measurements

Rhythm Classification  The key outcome measure was the classification of the first-recorded (presenting) rhythm at cardiac arrest as determined by EMS or an AED. For two cases in which the CEW was discharged in a hospital, we used physician classification.

The presenting rhythm was recorded from EMS reports as listed: asystole, “brady” (bradycardia), PEA, or VF. In some cases, an AED made the first rhythm classification of “shockable” or “nonshockable.” Generally, AEDs classify VF and wide-complex tachycardias with rates faster than 150–200 beats/min as “shockable” and other rhythms as “nonshockable.”¹⁵ A shockable rhythm was considered VF. In the five cases in which a rhythm was evaluated both by EMS and by an AED, we used the AED classification.

Whenever possible, the EMS and AED rhythm classifications were corroborated by review of ECG rhythm strips or response to rhythm-specific therapy. Resuscitation to a perfusing rhythm without defibrillation or antiarrhythmic drugs was considered evidence that the rhythm was not VF. Secondary rhythms recorded later in the resuscitation by EMS were analyzed only as a check on accuracy of EMS or AED rhythm diagnosis, not to classify the presenting rhythm.

Reconstruction of Time Course

Time From CEW Discharge to Collapse.  This interval was calculated from EMS system (9-1-1) logs, EMS field reports, and law enforcement records (including CEW electronic data logs and timelines in crime reports) whenever we had these data and could confirm synchronization of clocks (Figure 1). The remaining intervals were estimated from narratives of events. This interval was usually reported or could be estimated within 1 minute if it was 5 minutes or less. When clock times were not recorded, we used estimates based on the key events recorded by law enforcement: application of handcuffs, the duration of the struggle after handcuffs were applied, and whether the subject had been assessed and moved from the site of arrest or encounter after being handcuffed.

Figure 1.  Reconstruction of time course using EMS field report (A), law enforcement records (B), and EMS 9-1-1 system phone log (C). Field report (A) indicates that EMS was dispatched after first CEW discharge at 10:46. Accompanying narrative (not shown) states that, after arriving, EMS staged across the road, “out of the line of sight and gunfire.” Law enforcement reports include electronic log of last CEW discharges shown in (B), indicating the last CEW discharge at 10:49:28. Log displays total lifetime number of CEW discharges, local time, duration of each CEW discharge in seconds, temperature (°C), and percent battery capacity. EMS phone log (C) shows that law enforcement placed an urgent request for EMS at 10:59:13 when subject collapsed (first +). Field log (A) indicates EMS arrived on scene at 11:02 (second asterisk) and records the presenting rhythm as sinus bradycardia at 11:03 (third asterisk). Narrative (not shown) states, “Quick look pads show PEA (sinus brady).” In last entry in (C) at 11:03:10, law enforcement on scene confirms arrival of EMS for system operator. Synchronization of time codes indicated that the CEW clock was 5 minutes slow, so the final CEW application occurred at about 10:55. Thus the interval from the last CEW discharge to collapse was 4 minutes (10:55 to 10:59). The interval from collapse to presenting rhythm was also 4 minutes (10:59 to 11:03). Names of law enforcement personnel are redacted. CEW = conducted electrical weapon; ECG = electrocardiogram.

Time From Collapse to First Recorded Rhythm.  When EMS or law enforcement personnel who were already on scene with defibrillators recorded the presenting rhythm, we calculated or estimated this interval from time lines in law enforcement records, EMS field reports, incident reports of law enforcement officers, and EMS system (9-1-1) logs when available. This interval could usually be calculated or estimated within 1 minute when EMS was on scene or staged nearby at the time of collapse. In the remaining cases, the time from collapse to call for EMS—or call to EMS to collapse, whichever came first—was determined from law enforcement reports, and the time from the call until the rhythm was recorded was taken from EMS field reports and phone logs when available (see Figure 1).

A cardiac pathologist, who was blinded to presenting rhythm, reviewed autopsy data for presence and severity of heart disease. Heart weights were normalized for body height to determine the degree of hypertrophy¹⁶: Mild hypertrophy (grade 1) was 101%–125% of normal; moderate hypertrophy (grade 2) was 126%–150% of normal, and marked hypertrophy (grade 3) was more than 150% of normal. Coronary artery disease was mild if there was less than 50% cross-sectional luminal narrowing (grade 1), moderate if there was 51%–75% narrowing (grade 2), and severe if there was >75% narrowing (grade 3). Left-ventricular dilation and fibrosis were recorded as present or absent. The overall severity of heart disease (grades 0 to 3) was the sum of the degrees of hypertrophy and coronary artery disease up to maximum of 3. For example, if a patient had grade 2 hypertrophy and grade of coronary artery disease ≥1, the severity of heart disease was assigned grade 3.

Data Analysis

We used the chi-square test with Yates’ correction or the unpaired t-test to compare characteristics of study subjects and subjects who collapsed within 15 minutes but were not analyzed because of absence of rhythm data. Severity of cardiac disease was evaluated for correlation with the presenting rhythm using Fisher’s exact test.

Results

Subjects

All 422 subjects identified by any of the search terms used were found using the terms “TASER” and “death.”Figure 2 shows that we received records for 200 subjects who had nontraumatic sudden deaths after CEW discharges. Collapse occurred within 15 minutes in 118 subjects (59%). Of these, 62 subjects were excluded because EMS records of the presenting rhythm could not be obtained.

Figure 2.  Flow diagram shows selection of subjects. At each stage, number in bold above arrow denotes subjects progressing to next stage; number in parentheses below arrow denotes subjects who failed to meet the selection criterion. We requested records for 422 deaths linked to CEWs by media reports and received at least some records for 288 deaths (67% response rate). CEWs were not discharged in subduing 68 subjects, and the CEW probes did not deliver energy to one subject. Of the remaining 219 subjects, 19 died of trauma (blunt force trauma, 10; gun shot wounds, 8; drowning, 1). Thus, we received records for 200 subjects who had nontraumatic sudden deaths after CEW discharges. CEW = conducted electrical weapon.

The study population includes the remaining 56 subjects. Most were males who were using illegal drugs. The median age was 35 years. Table 1 shows that demographic characteristics and CEW electrode type did not differ significantly between the 56 study subjects and the 62 excluded subjects. The TASER CEW model was X-26 in 51 study subjects (91%) and 50 excluded subjects (81%), M-26 in 2 study subjects (4%), and unknown in 3 study subjects (5%) and 12 excluded subjects (19%).

Table 1.    Subjects with Nontraumatic Sudden Death After CEW Discharge Who Collapsed Within 15 Minutes (n = 118)

	Study Subjects (n = 56)	Excluded Subjects (n = 62)	p-value
CEW = conducted electrical weapon. *Overall, 45 study subjects took illegal drugs (80%) and seven took legal drugs (13%). Illegal drugs included cocaine, 24; methamphetamine, 15; phencyclidine (PCP), two; MDMA (ecstasy), three; ephedrine, two; lysergic acid diethylamide (LSD), one; and tetrahydrocannabinol (THC), five. Nine subjects took multiple illegal drugs. Legal drugs included olanzipine, two; mirtazapine, two; promethazine, one; bupropion, one. †Data were available for 60 excluded subjects; percentage of 60 is shown.
Age, yr (mean ± SD)	37 ± 10	36 ± 9	0.56
Sex (M:F)	54:2	61:1	0.46
Body mass index (mean ± SD)	30 ± 8	29 ± 6	0.42
Stimulants or psychoactive drugs	50 (89%)*	57 (95%)†	0.31
CEW electrode type
Probes	36 (64%)	47 (76%)	0.34
Drive-stun	12 (21%)	8 (13%)
Both	4 (7%)	2 (3%)
Unknown	5 (9%)	3 (5%)

Presenting Rhythms

Table 2 shows that the presenting rhythm was VF in four subjects (7%; 95% confidence interval [CI] = 3% to 17%) and not VF in 52 subjects (93%; 95% CI = 83% to 97%). The presenting rhythm was classified by EMS interpretation in 41 subjects (75%), AED algorithms in 13 subjects (23%), and physician interpretation in two subjects (4%). Figure 3 shows representative EMS records of the presenting rhythms.

Table 2.    Presenting Cardiac Arrest Rhythm as Classified by EMS (n = 56)

Rhythm	n (%)	n
AED = automatic external defibrillator; PEA = pulseless electrical activity; VF = ventricular fibrillation. *Values in parentheses indicate the number of ECG rhythms confirmed by the authors. †Two episodes of PEA were also classified by implantable cardiac defibrillator algorithms to not be VT or VF.
No VF
Asystole	52 (93)	21 (5)*
PEA ≥ 30 beats/min		11† (3)*
Bradycardia < 30 beats/min		5 (2)*
AED—no shock advised		11 (1)*
VF
VF	4 (7)	2
AED—shock advised		2 (1)*

Figure 3.  Representative EMS records. Each panel shows sequential entries with time in the left column. The column indicating cardiac rhythm (“ECG” or “Rhythm/Defib”) has black border added. The first entry in this column is the presenting rhythm. (A) Transition from perfusing sinus rhythm to PEA: a second black border denotes the blood pressure column (“B/P”). The presenting rhythm at 18:50 is sinus with a pulse of 76 beats/min and blood pressure of 99/48 mm Hg. Sequential entries indicate that blood pressure can no longer be measured and that the rhythm is irregular (“irreg”) at 18:55 and has a rate of 75–140 beats/min at 19:04. (B) Asystole. EMS record (top panel) and corresponding initial three-lead ECG rhythm strip (bottom panel). Chest compressions of cardiopulmonary resuscitation (CPR) are held at arrow. ECG = electrocardiogram.

Corroboration of Rhythm Classifications: ECG Recordings

We received ECG recordings of presenting rhythms for 12 of 56 subjects (21%). EMS personnel classified 10 of these rhythms (see Table 2, Figures 3B and 4); AEDs classified two of these rhythms, one as nonshockable (Figure 5A), and the other as shockable (Figure 5B). Four of seven presenting rhythms confirmed as asystole by ECG had identifiable QRS complexes with ventricular rates less than 5 beats/min (Figure 4).

Figure 4.  Sequential ECG rhythm strips during pharmacologic resuscitation from asystole. (A, 00:06:04) Profound bradycardia-asystole recorded when chest compressions were held just before the second dose of epinephrine and atropine. A single wide QRS complex is recorded at the left; artifact appears at right. (B, 00:06:34): Return of wide-complex rhythm at 43 beats/min and circulation after second doses of epinephrine and atropine. (C) Recorded 5 minutes later (00:11:18): irregular, mostly narrow-complex tachycardia without identifiable atrial activity, probably atrial fibrillation.

Figure 5.  Rhythms at cardiac arrest. (A and B) First recorded rhythms from AEDs with ECG rhythm strip and AED annotation channel. (A) Initial recording shows bradycardia. Sequential annotations indicate “Patient Connected” and “Initial Rhythm.” The AED algorithm then analyzed two rhythm segments and classified the rhythm showing profound bradycardia at 19 beats/min as “nonshockable.” (B) VF (subject 1). The initial recording (not shown) displays artifact from CPR. The tracing shows the first two segments analyzed, which the AED classifies as “shockable.” (C) Secondary EMS ECG tracing from VF subject 4, 10 minutes after initial record, which was not reviewed. Rhythm is shown immediately before and after defibrillation shock (delivered between dotted vertical lines highlighted by downward facing arrows). Preshock tracing shows an irregular, wide-complex tachycardia with isoelectric baseline and no identifiable atrial activity. Postshock, the rate, irregularity, and morphology of QRS complexes are indistinguishable from those in the preshock rhythm. Possible diagnoses include VT and conducted supraventricular arrhythmia with aberrancy (e.g., secondary to hyperkalemia). The lack of effect of defibrillation on the rhythm makes VT less likely. AED = automatic external defibrillator; CPR = cardiopulmonary resuscitation; ECG = electrocardiogram; VT = ventricular tachycardia.

Secondary rhythms recorded later in the resuscitation by EMS were available for an additional 11 subjects (20%). Of the 23 ECG tracings we reviewed (12 initial and 11 secondary tracings), we judged the diagnosis to be accurate in 22 (96%) and inaccurate in one secondary rhythm (4%). EMS classified this organized rhythm as VF, but we classified it as PEA (Figure 5C).

Corroboration of EMS Rhythm Classifications

Response to Treatment  Overall, 17 subjects were resuscitated to a perfusing rhythm. Sixteen subjects had return of circulation during pharmacologic resuscitation, excluding the diagnosis of VF (Figure 4). Presenting rhythm classifications in these subjects were asystole (n = 8), PEA (n = 4), bradycardia less than 30 beats/min (n = 1), and AED nonshockable (n = 3). One subject with an AED shockable rhythm was defibrillated. Overall, 14 of the 25 rhythms diagnosed as asystole by EMS were corroborated: seven by return of circulation during pharmacological resuscitation and 7 by review of the ECG.

Subjects with VF

Table 3 summarizes these subjects. For subject 1, who collapsed immediately (subject 6 in Table 4), neither drugs nor cardiac disease can be implicated; both the time course and the electrode location are consistent with electrically induced VF. In subject 2, neither is consistent. He was subdued without a CEW, but collapsed and became unresponsive while walking to the police car in handcuffs. To determine if he was feigning, the arresting officer delivered a 1 second drive-stun discharge to the calf. When the subject remained unresponsive, the officer recognized that he was in cardiac arrest. In subjects 3 and 4, electrode locations permit a current path through the heart, but the time from CEW discharge to collapse excludes immediate induction of VF. In both cases, subjects were using stimulant drugs, and subject 3 had cardiomyopathy. We received the initial ECG only for subject 1. In addition, we received a secondary tracing for subject 4, recorded 10 minutes after the presenting rhythm (Figure 5C). EMS diagnosed it incorrectly as VF.

Table 3.    Subjects in Whom Presenting Rhythm Was VF (n = 4)

Subject	Age/Sex	BMI	Details	ECG Reviewed	Time (min)*		Electrodes and No. of Discharges	Electrode Locations	Drugs	Cardiac Pathology
					Last CEW Discharge to Collapse	Loss of Pulse to ECG
BMI = body mass index; brady = bradycardia; CAD = coronary artery disease; CEW = conducted electrical weapon; ECG = electrocardiogram; EMS-D = EMS manual defibrillator; ICD = implantable cardioverter defibrillator; LE = law enforcement records; LSD = lysergic acid dimethylamide; VF = ventricular fibrillation. *Time intervals in parentheses are estimates. †Subsequent ECG diagnosed as VF shows organized tachycardia (Figure 5C).
1	25/M	28.4	Police AED	Yes	Immediate	(3–5)	Probes, 3 × 5 sec	Anterior chest: 1 midline lower chest, 1 superior to medial left clavicle	None	None
2	41/M	28.9	Police AED; “Shockable”; shocked to perfusing rhythm	No	Collapsed before CEW discharge	(3–5)	Drive stun, 1 × 1 sec	Right calf	Cocaine	Extensive anterior, septal, and inferior infarction; 3-vessel CAD (80–90% stenosis)
3	54/M	23.8	EMS-D; collapsed while under EMS care	No	8	2	Probes, 2 × 5 sec	Chest slightly to left of midline: 1 near level of nipple, 1 at costophrenic margin	Methamphetamine	Marked left-ventricular hypertrophy and enlargement; 50% left anterior descending coronary artery stenosis
4	18/M	19.7	EMS-D†	No†	5	6	Probes, 2 × 5 sec	Left chest slightly medial to nipple: 1 just below nipple and 1 near costophrenic margin	Methamphetamine, MDMA	None

Table 4.    Subjects in Whom Presenting ECG Record Was Reviewed (n = 12)

Subject	Age/Sex	BMI	Rhythm (ECG Type)	Time (min)*		Source Data for Time	Electrodes and No. of Discharges	Electrodes to Anterior Chest†	Drugs	Medical Conditions
				Last CEW Discharge to Collapse	Loss of Pulse to ECG
BMI = body mass index; brady = bradycardia; CAD = coronary artery disease; CEW = conducted electrical weapon; EMS-D = EMS manual defibrillator; ICD = implantable cardioverter defibrillator; LE = law enforcement records; LSD = lysergic acid dimethylamide; PEA = pulseless electrical activity; VF = ventricular fibrillation. *Time intervals in parentheses are estimates. If both probes and drive-stun modes were used, first interval is from last drive stun and second interval is from last probe discharge. †Number in parentheses is number of electrodes over anterior left chest.
1	22/M	18.7	Sinus brady 19 beats/min (EMS-D)	5	6	LE, EMS	Probes × 1	Yes (1)	Marijuana
2	47/M	28.4	Asystole (EMS-D)	7	0 (in ambulance)	LE, EMS video	Probe × 3	No	Cocaine, alcohol
3	34/M	30.0	PEA (hospital monitor)	5	0 (on monitor)	Medical record	Drive stun × 3	No	Cocaine
4	21/M	26.5	Asystole (EMS-D)	13	2 (EMS on scene)	LE, EMS video†	Probes and drive stun × multiple	No	LSD, MDMA
5	62/M	45.1	Sinus brady 18 beats/min (EMS-D)	(3–5), (3–5)	(2–3) EMS on scene	LE, EMS	Probes × 2; drive stun × multiple	No	None	Hypertensive cardiomyopathy; moderate CAD; heat exhaustion; schizophrenia
6	25/M	28.4	VF (police AED)	Immediate	(3–5) Police AED	LE	Probes × 3	Yes (2)	None
7	53/M	32.0	PEA (EMS-D; ICD interrogation)	(2–3)	(2–3) EMS on scene	LE, EMS	Probes × unknown	No	Methamphetamine	Dilated cardiomyopathy; ICD
8	40/M	31.5	Asystole (EMS-D)	1, (4–6)	8	LE, EMS, 9-1-1 log	Probes × multiple & drive stun × 1	No	Methamphetamine	Hypertensive cardiomyopathy
9	32/M	25.4	PEA (EMS-D)	(2–3)	(2–3) EMS on scene	LE, EMS	Probes × 5	No	Cocaine
10	17/M	21.1	Asystole (EMS-D)	(3–5)	(2–3) EMS on scene	LE, EMS	Probes × 2	No	None	Schizophrenia
11	34/M	34.4	PEA (EMS-D)	(8–12)	0 On Monitor	P EMS		No	Cocaine, marijuana
12	45/M	27.0	Asystole EMS ED	(1), (4–8)	(1–3) EMS on scene	P EMS	Probes × 1 and drive stun × 1	No	Cocaine, methamphetamine	Hypertensive cardiomyopathy; schizophrenia

Time From Last CEW Discharge to Collapse

Table 5A summarizes these time intervals for 56 study subjects. Calculated intervals are listed individually. The remaining intervals were estimated from narratives of the events. Subjects with electrically induced VF would be expected to collapse within 10 seconds. Overall, only five subjects collapsed in the first minute. Of these, one subject described previously collapsed immediately with VF (Figure 5B). Three of the four remaining subjects who collapsed in the first minute had non-VF rhythms. They were involved in protracted, grappling struggles with multiple law enforcement officers, who used multiple probe and drive-stun discharges as well as repeated applications of pepper spray and baton strikes until the subjects collapsed. Each subject collapsed more than 5 minutes after the last probe discharge, but within 1 minute of the last of more than 10 drive-stun discharges to the back or extremities. In all three subjects, non-VF rhythms were recorded within 5 minutes of collapse (EMS-diagnosed asystole, EMS-diagnosed PEA, and AED-diagnosed nonshockable rhythm). The fourth subject, who received only drive-stun discharges and had AED-diagnosed nonshockable rhythm, was resuscitated pharmacologically to sinus tachycardia, which is extremely unlikely in terminal asystole that degenerates from VF.

Table 5.    Time Intervals: Calculated From Reported Data or Estimated From Narratives of Events

Collapse Relative to Events	No. of Subjects	No. With Calculated Intervals	Calculated Intervals (min)	No. With Estimated Intervals	Estimated Intervals (min)
AED = automatic external defibrillator; CEW = conducted electrical weapon; ECG = electrocardiogram; NA = not applicable. *Six subjects were under EMS observation and two subjects were in hospital EDs under physician observation.
A. Time From Last CEW Discharge to Collapse
Immediate	1	1	0	0	NA
Before handcuffs	5	5	0.5, 1, 1, 1, 5	0	NA
After handcuffs, before being moved on site
Stopped struggling after  cuffs	8	5	2, 3, 3, 4, 7	3	2–3
Struggled for  several minutes	18	8	4, 4, 5, 6, 6, 6, 7, 8	10	3–5
Moved on site  before collapse	24	7	7, 8, 8, 10, 10, 12, 15	17	6–15
Total	56	26		30
B. Time from Collapse to First Recorded Rhythm
Subject on ECG monitor under observation*	8	8	0, 0, 0, 0, 0, 0, 0, 0	0	NA
EMS with defibrillator on scene, witnessed collapse	23	8	1, 1, 2, 2, 2, 2, 2, 2	15	1–3
Law enforcement used on-scene AED	4	1	3	3	3–5
EMS staged adjacent to scene	4	4	3, 3, 4, 4	0	NA
EMS in route	11	6	3, 5, 5, 6, 7, 8	5	Unknown
EMS called after collapse	6	2	5, 10	4	Unknown
Total	56	29		27

Time From Collapse to First Recorded Rhythm

Table 5B summarizes these intervals: immediate in eight subjects (14%) who were under observation on ECG monitors at the time of collapse, 3 minutes or less in 27 additional subjects (48%), 3 to 5 minutes in eight subjects (14%), 6 to 10 minutes in four subjects (7%), and unknown in nine subjects (16%). Overall, the interval was 3 minutes or less in 35 subjects (62%) and 5 minutes or less in 43 subjects (77%). The prevalence of short intervals is explained by two factors: 1) EMS is often dispatched before law enforcement because the subject’s erratic behavior suggests a psychiatric problem, and 2) law enforcement often calls for EMS at the onset of an altercation or after the first CEW deployment. None of the eight subjects who collapsed while on ECG monitors had VF as the presenting rhythm.

Cardiac Disease

Autopsy reports were available for all study subjects. Three autopsies indicated that subjects had been treated for heart disease, including two with nonischemic cardiomyopathy who had been treated with primary-prevention implantable cardioverter-defibrillators (ICDs).

Figure 6 summarizes postmortem findings. Seven subjects (14%) had marked hypertrophy, and one subject (2%) had severe coronary artery disease. Overall, heart disease was classified as marked in 14 subjects (25%) and moderate in 11 subjects (20%).

Figure 6.  Histogram showing heart disease found at autopsy. Gross fibrosis and left-ventricular (LV) dilation were recorded as present (+) or absent (−). Coronary artery disease, hypertrophy, and overall heart disease were graded on a scale of 1 to 3. PEA = pulseless electrical activity; VF = ventricular fibrillation.

Discussion

Our principal finding is that, in sudden death proximate to discharge of CEWs, VF is an uncommon presenting rhythm and collapse is rarely immediate.

Studies of Arrest-related Deaths

Sudden deaths of civilians temporally proximate to interactions with law enforcement were described before use of CEWs.^17,18 Most reported subjects were agitated young males, often chronic and acute users of stimulant or psychoactive drugs, who collapsed while being restrained.^19–23 A recent European publication confirms the international nature of this problem.²⁴

Initial studies focused on methods of restraining or subduing subjects,^17,23 providing few data regarding cardiac rhythms. Hick et al.²⁵ reported on cardiac arrests after restraint in five agitated and violent subjects not associated with use of CEWs. The presenting rhythm was asystole or PEA in four subjects and VF in one subject; all subjects had extreme metabolic acidosis. Stratton et al.¹⁹ reported 13 delirious subjects who died after being restrained. The rhythm was asystole or PEA in all three subjects who received CEW discharges and in nine of 10 subjects who did not receive discharges.

Cardiac Effects of CEWs

Because rapid cardiac electrical stimulation can cause cardiac arrest by inducing VF,^5,6 studies on sudden death after discharges of CEWs have focused on the risk of electrically induced VF.^8–13 Under some experimental conditions, CEWs can induce VF in animals using probe discharges across the anterior chest;^8,13,14 but they have not been reported to induce VF or any other arrhythmias in human volunteers.^26,27 There is uncertainty regarding how results of these studies relate to sudden death after tactical exposure to CEWs: VF thresholds for external electrical stimulation may differ between smaller, anesthetized animals and larger, awake humans. The controlled conditions of research in humans do not duplicate the stresses of tactical interactions, and volunteers, usually law enforcement officers, differ from decedents, who often use street drugs and are typically in less robust physical condition.

Present Study

To the best of our knowledge, this is the first systematic analysis of the presenting rhythm in sudden death proximate to use of CEWs. Of those collapsing within 15 minutes, the presenting rhythm was VF in only four of 56 study subjects (7%).

The time sequence and electrode location are both consistent with electrically induced VF in one subject (subject 1), and neither drug use nor cardiac disease provides alternative explanations. To the best of our knowledge, this is the first reported fatality suggestive of CEW-induced VF.

Conversely, both the time sequence and the electrode location are inconsistent with VF in subject 2. Electrode locations in subjects 3 and 4 could have placed the heart in the current pathway, but delays from CEW discharge to collapse exclude direct electrical induction of VF. Electrical induction of an intermediate arrhythmia, such as ventricular tachycardia (VT) or atrial fibrillation, is possible. However, electrical induction of stable VT that degenerates to VF is rare, especially in a structurally normal heart (subject 4),⁶ and degeneration of atrial fibrillation to VF is rare, except in unusual cases of ventricular preexcitation (Wolff-Parkinson-White syndrome).

Results in the Context of Prior Studies

The low incidence of VT/VF in our subjects is similar to that reported in small studies of subjects who were subdued without CEWs.^19,25 Our data confirm and expand the only known previous report of cardiac arrest rhythms associated with CEWs, in which Stratton reported no VF in three subjects.¹⁹ Our 12 subjects with initial ECG records is the largest group reported with validated rhythms in sudden deaths after CEW discharge. Our eight subjects who collapsed during ECG monitoring is the only group in whom the initial, cardiac arrest rhythm had been recorded after CEW discharges.

These results focus attention on the paradoxically high prevalence of asystole and PEA in nontraumatic sudden deaths after extreme stress or exertion during law enforcement interactions.²⁸ In combination with similar demographic and drug use profiles, they raise the possibility that the principal mechanism may be related to characteristics of the subject and the struggle. Metabolic acidosis and alterations of the central nervous system have been proposed, but not confirmed, as contributory factors. Experimental evidence does not suggest a mechanism by which CEWs contribute. Specifically, probe discharges cause little acidosis,²⁹ and drive-stun discharges should cause less. But we cannot exclude a CEW contribution by mechanisms yet unknown.

Limitations

Likelihood of Underdiagnosis of VF

Despite extensive efforts to obtain ECG records of the initial cardiac arrest rhythm, we succeeded in only 12 cases. It is unlikely many more ECGs could be obtained retrospectively, and unexpected sudden death after tactical discharges of CEWs is too rare for a prospective study. The largest known study³⁰ reported no atraumatic deaths within 15 minutes of CEW discharges in 996 consecutive subjects.

However, the fact that we reviewed ECGs in only 12 subjects requires us to consider the possibility that our methods failed to identify a significant number of subjects with electrically induced VF, because either EMS misinterpreted “fine” VF as “flatline” asystole or VF terminated before the initial ECG recording.

Likelihood of Misdiagnosis of VF as Asystole  Five lines of evidence suggest a low rate of misdiagnosis. First, misdiagnosis of VF as asystole is rare, either by EMS³¹ or AEDs.¹⁵ For example, emergency medical technicians analyzing a single ECG lead misdiagnosed VF as asystole in only three of 118 cardiac arrest patients.³¹ Second, the prevalence of VF is low in cases for which the EMS diagnosis is corroborated: of the 12 subjects for whom we reviewed ECG rhythm strips, VF was present in only one (8%). If we include subjects in whom an AED evaluated the rhythm (12 additional subjects), an ICD did not detect VF (one additional subject), and return of circulation occurred after pharmacologic resuscitation (10 additional subjects), the prevalence of VF was two in 35 (6%). Third, of the 23 total initial and subsequent ECG tracings reviewed, only one had a misdiagnosis; but in this case, VF was not underdiagnosed, but rather diagnosed in error (Figure 5C). Fourth, four of seven rhythms strips that were diagnosed as asystole and reviewed by the authors had at least one discernable QRS complex (Figure 4, top panel) rather than true flatline. Fifth, 14 of the 25 rhythms diagnosed as asystole by EMS were corroborated by ECG or return of circulation during pharmacologic resuscitation. Some of the remaining 11 episodes probably had QRS complexes, reducing the likelihood of misdiagnosing “fine” VF.

Likelihood of Presenting Rhythm Differing From Initial Rhythm  Of greatest concern is the likelihood that VF terminated before the rhythm was recorded. However, three considerations suggest that this is unlikely to have significantly altered our observed prevalence of VF.

First, in most subjects, the time from collapse to rhythm determination was short relative to the likely duration of VF. The duration of human VF required for degeneration to asystole has not been studied for ethical reasons. However, in pigs and dogs, electrically induced VF persists for at least 10 minutes.³² Rare cases in which ICDs have failed to terminate VF indicate that human VF can persist for more than 30 minutes.³³ Relative to the 10 minute time frame, the time from collapse to rhythm determination in sudden death after CEW discharge is short: 3 minutes or less in 62% of subjects and 5 minutes or less in 77%. This occurred because EMS usually was on scene, nearby, or en route prior to the subject’s collapse.

Second, if VF degenerated to asystole, we would expect a greater fraction of VF episodes to degenerate with longer intervals from collapse to rhythm determination, resulting in an inverse correlation between the interval and the prevalence of VF. But the reverse occurred: VF was recorded in none of the eight subjects who had cardiac arrests during continuous ECG monitoring, one of the remaining 27 subjects (4%) whose rhythms were recorded in the first 3 minutes after collapse, and three of 21 subjects (14%) in whom the time from collapse to rhythm determination was greater than 3 minutes or unknown.

Third, bradycardia or asystole may have degenerated to VF before the rhythm was recorded, offsetting any degeneration of VF to asystole. Although VF lasts for minutes before degeneration to asystole, a single long pause usually precipitates bradycardia-dependent VF.^34,35 This must be considered given the inverse relationship between the time from collapse and prevalence of VF.

Time Course and Current Pathway in Electrically Induced VF  Any estimate of the likelihood of underdiagnosing VF must consider both the time course, which is typically immediate for electrically induced VF,⁶ and the electrode type and location, because CEW discharges in animals have induced VF only using probe electrodes over the anterior chest.^8–13 These considerations lead us to focus on subjects who collapsed with non-VF rhythms rapidly after CEW discharges through probe electrodes, but the only subject who collapsed in the first minute after a probe CEW discharge had VF. Drive-stun discharges have not been reported to induce VF in animals, even when delivered directly over the heart.¹¹ Thus the low detected prevalence of VF in our study is consistent with the observation that, in 98% of subjects, the time course or current pathway is atypical of electrically induced VF in animal experiments.

Method of Identifying Subjects

The Internet search method has been used infrequently in medical research³⁶ because confidentiality of patient data is protected. We could use it in this study because sudden deaths after CEW discharges attract media attention, and media reports often include the names of decedents.

Our method identified only those subjects whose deaths were reported publically and detected by the search terms used. Once subjects were identified, our access to data depended on cooperation of different law enforcement and EMS agencies with the Freedom of Information Act. A national database of in-custody sudden deaths would facilitate identifying subjects and reduce sample bias inherent in our method.

We do not know if our study subjects are representative of those who die suddenly after CEW discharges. However, our subjects had demographic and drug-use profiles similar to those of subjects we excluded because of insufficient data, subjects identified by prospective study of nonfatal use of CEWs by law enforcement,^30,37 and subjects in prior reports of sudden deaths during law enforcement interactions—either associated²¹ or unassociated with use of CEWs.^18,19,24 At autopsy, the overall incidence of moderate and marked heart disease was 45%, not statistically different from the 54% reported for subjects who died within 24 hours of CEW discharge²¹ (p = 0.50). These considerations suggest that our subjects are in fact representative, but do not exclude undetected bias.

Our methods did not identify subjects who were resuscitated from cardiac arrests and survived to be discharged from the hospital. However, the fraction of subjects who were resuscitated to a perfusing rhythm was not statistically different for VF and non-VF presenting rhythms: one of four subjects (25%) versus 16 of 52 subjects (31%; p = 1.00). This suggests that rhythms in subjects who survived to be discharged and those who died would have a similar distribution.

In summary, study subjects represent only a small fraction of sudden deaths proximate to CEW discharges. Although the method of selecting them introduced bias, there is no evidence that this bias influenced the distribution of presenting rhythms; however, we cannot exclude this possibility.

Conclusions

In sudden deaths after conducted electrical weapon discharge, collapse is rarely immediate, and VF is an uncommon presenting rhythm. Only one death was suggestive of electrically induced VF. Except for the eight subjects who collapsed during electrocardiogram monitoring, our estimates of initial cardiac arrest rhythms are subject to the limitations discussed, and we do not know if selection bias limits extrapolation of our findings to subjects identified by another strategy. With these qualifications, our data do not support electrically induced VF as a common mechanism of sudden death after conducted electrical weapon discharge. The specific mechanisms for most of these deaths remain unknown.