In the United States in 2020, an estimated 45,979 people died by suicide and 24,576 by homicide.1,2 Some healthcare workers' deaths are attributable to both types of violence. Concerns about burnout, depression, and suicide among healthcare professionals have garnered increasing attention.3,4 Healthcare workplace violence also has received increasing attention and is a major safety concern in healthcare settings.5,6 The Bureau of Labor Statistics (BLS) has warned of the public health challenges of increasing violence in healthcare.7Violence in the workplace takes many forms, including fatal attacks on health professionals.8

Increasingly comprehensive data have been emerging about the violent deaths of healthcare personnel since the CDC's National Violent Death Reporting System (NVDRS) was created in 2003. Its database of deaths by suicide and homicide has been interrogated for deaths of physicians, nurses, pharmacists, psychologists, and other healthcare professionals.9-14 Studies have summarized decedents' demographics, methods of suicide, and homicide victims' relationships with assailants.

The incidence of healthcare professionals who die by suicide has been estimated by some researchers to exceed rates in the general population.9 Consistent with patterns of violent death in the general population, healthcare workers are more likely to die by suicide than homicide.15 A systematic review revealed that physicians are at risk for death by suicide and emphasized the need to implement studies investigating death by suicide in other healthcare professionals.16

The rate of homicides of healthcare professionals also is alarming. NVDRS data between 2003 and 2018 revealed that 56 physicians and 13 psychologists died by homicide.12,13 Robiner and colleagues analyzed NVDRS homicide data for 10 types of healthcare professionals, summarizing demographic data and calculating homicide rates in those professions' workforces.17 Homicides of healthcare professionals are not limited to the United States. For example, more than half of 21 physician homicides in Italy were perpetrated by patients and their relatives.18

Such reports raise questions about the extent and characteristics of violent deaths of physician associates/assistants (PAs). Burnout, depression, and suicidal ideation have been reported in PA trainees and faculty.19,20PA students have reported higher anxiety and depression than the general population.21 Our literature search using the terms physician assistant, homicide, suicide, kill, murder, and death identified no previous studies of PAs' violent deaths. However, PA homicides have been reported in the media.22,23 To bridge this gap in the literature, enhance understanding of the epidemiology of PAs' violent deaths, and raise awareness about them, we analyzed NVDRS data on PAs who died by suicide or homicide. This study sought to delineate the incidence of PAs' violent deaths, their demographics and associated characteristics, and to establish a baseline for future longitudinal research.

The NVDRS was developed by the CDC in 2003 and incorporates data from coroners' reports, police reports, and death certificates. Researchers can use these data to develop initiatives to prevent violent deaths and to assess the effects of those initiatives. The system provides demographic data and other information about decedents (for example, factors involved in incidents, toxicology reports, methods of attack, history of personal problems) and, where known, about their relationships with assailants.

The number of jurisdictions participating in the NVDRS has increased from 7 states to 50 states plus Puerto Rico and the District of Columbia, but only 48 provided sufficient data to be included in the 2020 NVDRS dataset. Figure 1 presents the number of states reporting to the NVDRS. Incomplete data in two states in 2020 and other states in other years limited reporting. Using US Census Bureau data, we determined that the incomplete 2020 NVDRS data captured 85% of the US population.24 Assuming that reporting and nonreporting areas had similar rates of suicide and homicide allowed calculation of extrapolated national estimates of PA deaths by suicide or homicide for 2020. These in turn were compared with BLS estimates of the PA workforce to derive estimated rates of suicide and homicide among PAs in 2020.25

Cumulative PA homicides and suicides and state participation in the NVDRS The completeness of data varied over years as the NVDRS evolved. Enrollment in the program was staggered, and some states submitted incomplete records at times. For example, in some years, Illinois and Pennsylvania collected data in a subset of counties representing only a portion of violent deaths in those states. Other examples include New York's exclusion in 2019 and Hawaii's exclusion for 3 years due to incomplete reporting. California only participated for 4 years, and in those years, only some counties participated; Texas participated only in 2020, reporting data for four participating counties. Florida has only recently participated, but data have not been sufficient for inclusion in the dataset. Consequently, the numbers presented underestimate the true incidence of violent deaths.

The NVDRS was queried for anonymized demographic and situational data about PA deaths by suicide and homicide between 2003 and 2020. Tables were created based on descriptive data in electronic spreadsheets. For the purpose of this study, both death by suicide and death by suicide following homicide committed by the PA were coded as suicide. Both authors reviewed recorded occupational classifications and the narratives of law enforcement and health departments provided in the NVDRS data to reach consensus in identifying PAs using the search terms physician assistant and physician asst. The study was deemed exempt by the University of Minnesota institutional review board.

Violent deaths of PAs were reported every year since 2004, the second year of the database, totaling 102 through 2020. Figure 1 presents the cumulative numbers of PA suicides and homicides. The 93 suicides and 9 homicides reveal an incidence of suicide nearly 10 times that of homicide (Figure 2). Three PAs died by suicide after having committed homicides. One death by suicide followed the PA's partner's death by suicide.

PA deaths by suicide or homicide, NVDRS 2003-2020

Table 1 summarizes demographics of PAs who died violent deaths. Overall, modal age range was 50 to 59 years. The 67 males' age (M = 52.8 years) exceeded the 35 females' age (M = 43.7 years). Most (n = 91; 89.2%) were White, non-Hispanic; Black and Asian PAs accounted for 3.9% and 4.9%, respectively. Most (n = 61; 59.9%) were married at the time of their death, 18 were divorced, 18 had never married, and 5 were widowed.

A plurality of PAs who died by suicide (43%) and a majority of those who died by homicide (77.8%) were in the South (Table 2). A quarter (24.7%) of the PAs who died by suicide, but none who died by homicide, were in the West. The Midwest and Northeast had fewer incidents, accounting collectively for nearly a third (31.4%) of PA violent deaths.

Most (72.5%) violent deaths were in PAs' homes including 55.6% of the deaths by homicide and 74.2% of those by suicide (Table 3). No homicides were in work settings; however, at least four deaths by suicide were in work settings. Locations were not reported for four deaths by suicide.

A plurality (n = 42; 41.2%) of deaths involved firearms (Table 4). Guns were involved in a greater proportion of deaths by suicide (n = 40; 43%) than homicide (n = 2; 22.2%). The frequency of gun-mediated death by suicide was 38% greater than the next most common method, overdose or intoxication (26.9%). Half (n = 32, 50.8%) of men who died by suicide used firearms, compared with about a quarter (n = 8, 26.7%) of women.