Physiological Attributes of Arrest-Related Sudden Deaths Proximate to the Application of TASER Electronic Control Devices: An Evidence Based Study of the Theory of High-Risk Groups
Date
2013-11
Authors
Williams, Howard E.
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
TASER electronic control devices (ECDs), manufactured by TASER International, Inc. in Scottsdale, Arizona, have become a popular tool for law enforcement. TASER International has sold more than 710,000 devices to 16,880 agencies in 107 countries. Although other manufacturers produce comparable types of electro-shock weapons, TASER products are the most commonly used in the United States and worldwide.
Unfortunately, more than 870 people worldwide have died unexpectedly following law enforcement officers’ uses of TASER ECDs. Currently, there is no research definitively establishing a causal relationship between the use of an ECD and the death of a person exposed to it. However, some recent studies suggest that application of TASER technology is responsible for sudden unexpected deaths. The ever increasing number of deaths following application of TASER ECDs and the growing number of cases wherein a coroner or medical examiner attribute the use of an ECD as a cause of death or as a significant contributing factor to the death raise legitimate concerns about the safety threshold of the devices.
Researchers have proposed and tested many theories of why people die following the application of ECDs, including direct electro-stimulation of cardiac muscle, interference with breathing, and metabolic changes resulting in acidosis. Thus far, human model experiments have produced no evidence to support these theories. Another theory, which has recently appeared in the literature, has received no empirical testing—the theory of high-risk groups. High-risk group theory postulates that elderly people, young children, people with pre-existing cardiovascular disease, people with pacemakers and implantable cardioverter-defibrillators, people under the influence of drugs (amphetamines, cocaine, lysergic acid diethylamide, marijuana, opiates, and/or phencyclidine) or with a history of drug abuse, people intoxicated from alcohol or with a history of chronic alcohol abuse, people under extreme psychological distress or who exhibit signs of excited delirium, people who are mentally ill or taking psychotropic medications, people subjected to repeated or multiple applications, and pregnant women are at a heightened risk of serious injury or death following application of a TASER ECD.
What the current literature fails to consider is that the same physiological attributes that are presumed to render members of high-risk groups more vulnerable to serious injury or death following application of a TASER ECD might render these same people more vulnerable to serious injury or death regardless of the tactics or weapons that officers use to subdue them. If that hypothesis is correct, the use of TASER ECDs on people in high-risk groups might be irrelevant to arrest-related sudden deaths.
The potential for fatal adverse effects on high-risk groups when using other less lethal tactics and methods versus the potential for fatal adverse effects on high-risk groups following the use of a TASER ECD is currently unknown. Thus far, research has not directly addressed the question. By examining autopsy and toxicology reports of the deceased and comparing differences in the physiological attributes of arrest-related sudden deaths, one can then estimate whether a difference exists in high-risk group attributes between deaths proximate to the use of a TASER ECD and deaths not involving the use of an ECD.
An arrest-related sudden death is a death that occurs following a collapse within 24 hours after the initial arrest or detention. The death must be unexpected, must not be the result of trauma or injury that a layperson could readily discern needs medical attention, and must follow a sudden change in clinical condition or the beginning of symptoms from which the deceased does not recover. It does not include police shootings and suicides.
This work was a retrospective open source research study of publicly available autopsy and toxicology reports designed to compare the physiological attributes of high-risk group theory to two groups of arrest-related sudden deaths, TASER ECD-proximate deaths and non-ECD deaths. A non-proportional stratified random sample of 300 publicly available reports was obtained for study and coded for physiological attributes of high-risk groups. Descriptive and inferential quantitative statistics, specifically Student’s t-test, Pearson’s chi-square, and bivariate logistic regression (logit) were used to analyze the data and compare attributes for any significant differences observed. Additionally, crisp set Qualitative Comparative Analysis (csQCA) was used to examine whether any configuration of causal conditions described a unique path to arrest-related sudden death following application of a TASER ECD.
No statistically significant difference in the frequencies of non-ECD deaths and ECD-proximate deaths was observed in several variables of high-risk group theory: cardiovascular disease, presence of drugs, history of drug abuse, alcohol present, and presence of psychotropic medications. ECD-proximate deaths were more likely than non-ECD deaths in cases involving excited delirium. ECD-proximate deaths were more likely than non-ECD deaths in cases involving cocaine and in cases involving a history of chronic alcohol abuse. Insufficient data existed to calculate any difference in frequencies of deaths involving pacemakers and implantable cardioverter-defibrillators. The two remaining tenets of high-risk group theory, multiple applications and pregnancy, were not examined in this study.
Examined independently with a Student’s t-test, people in the sample who suffer an arrest-related sudden death following application of a TASER ECD are approximately 2.67 years younger than those people who suffer an arrest-related sudden death absent the use of a TASER ECD, that difference being statistically significant (df = 298, t = 2.13, p = 0.03). However, in two logit models, while controlling for other predictive variables, the difference in ages between the groups is insufficiently large to establish statistical significance (z = -1.33 and z = -1.43).
Sex data from the sample indicates that arrest-related sudden deaths following application of a TASER ECD more frequently involve males than females, but the effect size is small (df = 1, χ2 = 6.80, p < 0.01, φ = -0.15). In the logit models, while controlling for the other predictive variables, the difference in sex in the sample is insufficiently large to establish statistical significance (z = 1.77 and z = 1.75).
Examined by χ2 analysis using four categorical designations for race, the differences in frequency of an arrest-related sudden death following application of a TASER ECD compared to deaths not involving an ECD for Whites, Blacks, Hispanics, and other races/ethnicities is statistically insignificant ( df = 3, χ2 = 1.10, p = 0.78, Cramér’s V = 0.06). When designated as a dichotomous variable of White = 0 and all other races/ethnicities = 1, the difference in race remains statistically insignificant in both logit models (z = -0.17 and z = 0.18).
Analyzed through χ2, the difference in arrest-related sudden deaths following application of a TASER ECD compared to deaths not involving an ECD is statistically insignificant for people with cardiovascular diseases (df = 1, χ2 = 0.86, p = 0.35, φ = 0.05). In the logit models, the difference observed in cardiovascular disease is insufficiently large to establish statistical significance (z = -0.26 and z = -0.11). Insufficient data exist to analyze the significance of the pacemakers and ICD variable. Viewed through the significant configurations in csQCA, cardiovascular disease is associated more frequently with non-ECD deaths than with ECD-proximate deaths. Only when cardiovascular disease interacts with ExDS does it become more frequent in ECD-proximate deaths, but that association is more likely due to the influence of ExDS than of cardiovascular disease.
When calculated with χ2, the difference in frequencies between arrest-related sudden deaths involving drugs following application of TASER ECDs compared to deaths not involving an ECD is insufficiently large to establish statistical significance (df = 1, χ2 = 2.05, p = 0.15, φ = 0.08). That relationship holds true for individual analyses of amphetamine (df = 1, χ2 = 0.02, p = 0.88, φ = 0.01), cocaine (df = 1, χ2 = 2.63, p = 0.11, φ = 0.09), marijuana (df = 1, χ2 = 0.02, p = 0.88, φ = -0.01), opiates (df = 1, χ2 = 2.01, p = 0.16, φ = 0.09), and PCP (df = 1, χ2 = 2.01, p = 0.16, φ = -0.08). Insufficient data exist to analyze LSD individually.
In logit model one, while controlling for the other predictive variables, the difference in drugs present is insufficiently large to establish statistical significance (z = -1.92). In model two, however, the difference in drugs is significant (z = -2.06). When viewed through the significant configurations in csQCA, drugs are associated more frequently with non-ECD deaths than with ECD-proximate deaths. Only when drugs interact with ExDS and cardiovascular disease does the variable become more frequent in ECD-proximate deaths, but that association is more likely due to the influence of ExDS than of drugs.
Examined in χ2, the difference in frequencies between arrest-related sudden deaths for people with a history of chronic drug abuse compared to deaths not involving a history of chronic drug abuse is insufficiently large to establish statistical significance (df = 1, χ2 = 0.81, p = 0.37, φ = -0.03). The difference in history of chronic drug abuse was also insufficiently large to establish statistical significance in the logit models (z = 0.55 and z = 0.47).
When examined in isolation by χ2, the difference in frequencies between arrest-related sudden deaths for people exhibiting excited delirium compared to people without excited delirium is sufficiently large to establish statistical significance (df = 1, χ2 = 16.79, p < 0.01, φ = 0.24), and both logit models calculate a statistically significant difference (z = 3.01 and z = 3.37). Viewed through csQCA, only three significant configurations are more likely to appear in ECD-proximate deaths than in non-ECD deaths. One of those three configurations contains no variable of high-risk group theory. The other two contain ExDS, and none of the significant configurations that are more likely to appear in non-ECD deaths contains ExDS. Arrest-related deaths involving ExDS are more likely following application of a TASER ECD than they are in non-ECD cases.
Examined in isolation by χ2, the difference in frequencies between arrest-related sudden deaths for people with alcohol in their systems compared to people without alcohol is insufficiently large to establish statistical significance (df = 1, χ2 = 0.02, p = 0.89, φ = 0.01). The logit models also calculate a statistically insignificant difference (z = 0.26, z = 0.17, and z = 0.52). Viewed through csQCA, alcohol does not appear in any of the substantial configurations that appear more frequently in ECD-proximate deaths, although it does appear in the one configuration that occurs with equal frequency in both groups.
When examined in isolation by χ2, the difference in frequencies between arrest-related sudden deaths for people with a history of chronic alcohol abuse compared to people without a history of chronic alcohol abuse is sufficiently large to establish statistical significance (df = 1, χ2 = 9.61, p < 0.01, φ = 0.18), meaning ECD-proximate deaths are less likely than non-ECD deaths to involve people with a history of alcohol abuse. Both logit models calculate a statistically significant difference (z = -2.43 and z = -2.37). History of chronic alcohol abuse does not appear in the significant configurations of csQCA.
When examined in isolation by χ2, the difference in frequencies between arrest-related sudden deaths for people with mental illness compared to people without a mental illness is sufficiently large to establish statistical significance (df = 1, χ2 = 13.78, p < 0.01, φ = 0.21), meaning ECD-proximate deaths are more likely than non-ECD deaths to involve people with mental illness. Logit model one also calculates a statistically significant difference (z = 2.81). However, the mental illness and psychotropic medications variables are collinear (adjusted LR χ2(1) =. 100.24). Combining the two variables into one variable for logit model two, the difference becomes insufficiently large to establish statistical significance (z = -1.43). Mental illness does not appear in the significant configurations of csQCA.
When examined in isolation by χ2, the difference in frequencies between arrest-related sudden deaths for people taking psychotropic medications compared to people without psychotropic medications is insufficiently large to establish statistical significance (df = 1, χ2 = 0.24, p = 0.63, φ = 0.03). Logit model one also calculates a statistically insignificant difference (z = -1.86).
In the csQCA truth table, 24 case configurations emerge from the 32 causal configurations possible for five dichotomous variables. The csQCA truth table cannot be resolved of conflicting configurations, so prime implicants cannot be deduced. Of the eight significant configurations with frequency cell counts of ten or more, all are conflicted. Only three of the eight significant configurations are more likely to be involved in an ECD-proximate death than they are to be involved in a non-ECD death. One of those configurations contains no condition germane to high-risk group theory. The other two configurations contain ExDS in combination with drugs or with drugs and cardiovascular disease. Cardiovascular disease and drugs do not appear in any configuration more likely to be involved in an ECD-proximate death, except for configurations that interact with ExDS. Mental illness does not appear in any of the eight significant configurations.
These results have significant policy and legal consequences. Many law enforcement agencies are re-examining their policies on the use of ECDs on people with pre-existing cardiovascular disease, on people intoxicated on drugs, on people who are intoxicated on alcohol, or on people with mental illness will have little effect in reducing arrest-related sudden deaths. Conversely, the data indicate that use of an ECD reduces the frequency of arrest-related sudden deaths for people intoxicated on cocaine or who have a history of chronic alcohol abuse. The literature indicates that the use of an ECD reduces injuries to officers and suspects. Consequently, policies restricting the use of ECDs will likely increase non-fatal injuries to officers and suspects without the concomitant benefit of reducing deaths. The results related to ExDS, however, do indicate an increased frequency of death following application of an ECD. Further research is necessary to clarify why, but, in the meantime, it might prove prudent for officers to consider using other force options, whenever possible, on subjects who display the constellation of symptoms related to ExDS.
The courts have forewarned that they will consider in deciding the objective reasonableness of a use of force whether an individual is emotionally disturbed, suffering from a mental illness, or suffering from diminished capacity, which can include severe intoxication, drug abuse, a discernible mental illness, or any other condition apparent to the officer that would make that use of force likely to result in significant risk to the individual’s health or well-being. The results in this study indicate that the courts should consider that the use of an ECD is no more likely than other force options to result in a significantly increased risk to an individual’s health or well-being. In cases of intoxication from drugs, particularly intoxication from cocaine, the use of an ECD is less likely to result in a significantly increased risk to an individual’s health or well-being. On the other hand, there is an indication that the use of an ECD might result in a significantly increased risk to an individual’s health or well-being if that individual is experiencing ExDS, thereby increasing potential liability exposure.
Description
Keywords
Arrest-related death, Cardiovascular disease, Custody death, Electronic control device, Excited delirium, Mental illness, Sudden-death, TASER, Use of force
Citation
Williams, H. E. (2013). <i>Physiological attributes of arrest-related sudden deaths proximate to the application of TASER electronic control devices: An evidence based study of the theory of high-risk groups</i> (Unpublished dissertation). Texas State University, San Marcos, Texas.