Abstract
Aging adults are at risk for multiple types of abuse including emotional, sexual, or physical abuse. Adults in skilled nursing or assisted living facilities might be particularly vulnerable to injury, including fatal injury resulting from incidents of resident-to-resident aggression (RRA). The purpose of this research is to report findings from the multistate National Violent Death Reporting System (NVDRS) related to fatal RRA incidents between years 2003 and 2016. We analyzed structured categorical data and unstructured narrative data on 101 identified fatal RRA incidents among individuals aged 65+. Findings included that victims average 16 years older than exhibitors and dementia diagnoses are present in most cases. Qualitative analysis revealed that events are described as both unexpected and preventable. Training to improve long-term care staff recognition and reporting of RRA events may potentially improve prevention and provide more accurate trend data. NVDRS data provide a useful source to follow trends in fatal RRA going forward.
Keywords
Introduction
The proportion of the U.S. population aged over 65 is expected to increase such that by the year 2030, 20% of all U.S. residents will be 65 or older (U.S. Census Bureau, 2018). According to the National Institute on Aging (2016), roughly 10% of adults over age 60 experience some type of abuse including emotional, sexual, or physical abuse. Older adults who live in skilled nursing and assisted living facilities may be at particular risk for injury, including fatal injury, due to physical frailty. Cognitive impairments, at times including dementia, also present a risk factor for abuse due to interference with individuals’ judgment and demonstrated association with aggressive behaviors (Dettmore, Kolanowski, & Boustani, 2009).
According to Gimm, Chowdhury, and Castle (2018), approximately 7.6% of all assisted living facility residents in the United States have demonstrated physical aggression toward staff or other residents. Caspi (2018) used media reports along with analysis of death records from the Ontario, Canada, Geriatric and Long Term Care Review Committee (GLTCRC) to identify 105 deaths worldwide, including 42 in the United States, that resulted from incidents of resident-to-resident aggression (RRA) occurring in long-term care (LTC) facilities between 1988 and 2017 and observed that lack of a centralized, comprehensive system for reporting “injurious and fatal [RRA]” (p. 301) presents a substantial challenge for researchers and practitioners who hope to better understand trends and correlates of RRA and aspire to develop interventions or programs to reduce likelihood and potential for injury due to RRA.
The purpose of this research report is to present findings from a data source not used thus far by researchers to explore fatal incidents of RRA. The National Violent Death Reporting System (NVDRS) is a multistate database implemented by the U.S. Centers for Disease Control and Prevention to track contextual details of homicides and suicides.
Previous RRA Research
Following publication of a seminal work by Shinoda-Tagawa and colleagues in 2004, who integrated public health data from Massachusetts with Centers for Medicare and Medicaid Services reports, U.S.-based researchers have published articles exploring RRA that reflect a variety of designs. Pillemer et al. (2011); Gimm et al. (2018); Lachs, Bachman, Williams, and O’Leary (2007); and Lindner et al. (2007) used secondary analysis of existing records. Rosen et al. (2008a), Rosen et al. (2016), Schonfeld (2003), and Snellgrove, Beck, Green, and McSweeney (2013) used field observation methods, interviews, or surveys of staff and residents. Teresi et al. (2013) reported results from an education-based intervention program. Ferrah et al. (2015) and Rosen, Pillemer, and Lachs (2008b) conducted reviews of prior literature. As described above, Caspi (2018) assessed homicides from RRA as reported in media and Ontario, Canada, government reports. Gimm et al. and Schonfeld focused on RRA in assisted living facilities and Caspi included reports from any type of facility; all other reports focused on RRA in skilled nursing facilities.
Most authors who examined multiple types of aggressive incidents (i.e., Gimm et al., 2018; Lachs et al., 2016; Rosen et al., 2008a; Schonfeld, 2003) found that verbal aggression occurred more frequently than physical aggression, although Rosen et al. (2008b) presented conflicting information regarding proportion of verbal versus physical incidents, depending on source of information. Based on surveys of assisted living facility staff and administrators, Schonfeld (2003) concluded that only 3.6% of all reported incidents involved verbal aggression between or among residents, and no incidents of physical aggression directed toward other residents were disclosed.
Findings from multiple studies (Caspi, 2018; Gimm et al., 2018; Shinoda-Tagawa et al., 2004) suggested that primary correlates of RRA included cognitive impairment such as dementia or mental illness. Gimm et al. (2018) reported that males were more likely than females to be exhibitors although Shinoda-Tagawa et al. (2004) reported greater likelihood of being injured for male nursing home residents. Caspi (2018) reported that “exhibitors” were more likely to be male and be younger than “targets” (p. 292).
Common precipitating incidents included intentional or accidental intrusion of another’s space, theft of possessions, disagreements with roommates, and conflicts over shared resources (Ferrah et al., 2015; Lachs et al., 2016; Shinoda-Tagawa et al., 2004; Snellgrove et al., 2013). Staff responses to RRA included physical or verbal interventions separating individuals, changing rooms, and notifying other staff (Rosen et al., 2016; Rosen et al., 2008b). Rosen et al. (2016) interviewed Certified Nursing Assistants and concluded that incidents were infrequently reported or documented. Teresi et al. (2013) implemented a staff education intervention designed to improve recognition and increase reporting of RRA. Results suggested substantial increases in number of reported incidents, with 819 RRA versus 102 incidents reported during two 6-month time periods by treatment versus control units.
Caspi (2018) was one of only two U.S. authors to focus on fatal incidents. Lindner et al. (2007) analyzed Allegheny County, Pennsylvania, coroner’s office reports reflecting the years 1993 to 2003, and identified only two deaths from homicide from a total of 207 autopsied incidents. One of the identified homicides resulted from a resident being pushed by another resident.
NVDRS
The NVDRS is a multistate database managed by the U.S. Centers for Disease Control and Prevention (CDC, 2019) to systematically track details regarding violent deaths and suicides throughout the 50 U.S. states (Paulozzi, Mercy, Frazier, & Annest, 2004). The number of reporting states has increased such that as of 2018, 50 states, Washington, D.C., and Puerto Rico, reported at least a proportion of homicides and suicides (CDC, 2019).
The NVDRS database tracks demographic, toxicological, contextual, and other available detail derived by data abstractors from coroner and medical examiner, law enforcement, and toxicology reports, and death certificates (Barber et al., 2016). De-identified case level data are available via the NVDRS-restricted access database (RAD; CDC, 2017). Available RAD data include a range of factors, and abstractor-produced summaries of source narrative documents. Although NVDRS cases do not specifically identify RRA, researchers are able to identify and verify cases through use of relevant criteria including age of victim and facility type where injury occurred and thoughtful review of narrative reports. The narrative segments of NVDRS data are also appropriate sources for in depth analysis processes of unstructured data, such as qualitative open coding.
Prior researchers have contributed substantial insight into correlates and precursors of RRA in the United States although the most recently used national data in a research report (Gimm et al., 2018) reflected reporting year 2010. Caspi (2018) included reports as recent as 2017 but acknowledged that reliance on media and one government source limited potential to present complete data. Several reports (e.g., Lachs et al., 2016; Pillemer et al., 2011; Rosen et al., 2008a; Rosen et al., 2016; Rosen et al., 2008b; Teresi et al., 2013) took place in New York with others (e.g., Lachs et al., 2007; Shinoda-Tagawa, 2004) focusing on other states within the Northeast United States. Therefore, the extent to which identified trends are stable through time and geography is presently unknown. Along with this, data from NVDRS offer some particular advantages for analysis of RRA. First, NVDRS homicide records might include detailed demographic and diagnostic information regarding both victim and exhibitor. Second, because the primary sources have their origin in law enforcement and medical examiner reports, NVDRS reports provide more consistent and objective information than is available through subjective means such as self-report data collection or media report review. A related advantage is that identified NVDRS records include only the population of interest rather than a convenience or opportunistic sample. Finally, as noted previously, the nature of NVDRS records that include categorical and narrative data facilitates analysis of both structured and unstructured, or qualitative, data.
Clearly, analysis of NVDRS data has potential to contribute additional insight and understanding to the emerging public health problem of RRA. Therefore, the purpose of this article is to report findings from analysis of both structured and unstructured NVDRS data, to consider to what extent our findings confirm or clarify findings from previous research reports, to provide recommendations for ongoing analysis of RRA trends, and to provide insight which may be useful in the development of staff training programs or interventions that might reduce the impact of RRA.
Method
We requested NVDRS RAD data for violent deaths occurring with a victim over the age of 65, for all available years (2003-2016), which included 3,860 total cases. States reported all cases with the exception of Illinois, Pennsylvania, and Washington; by agreement with the CDC these states reported at minimum 80% of cases. We utilized the narrative descriptions of events to identify all homicides that occurred in an LTC facility, where both the victim and exhibitor were a resident, and extracted 122 cases for further review. Following the secondary review, we removed 21 cases. Cases were excluded if they occurred in another type of facility (for example a substance abuse treatment facility or hospital), if either the patient or exhibitor was not a resident of the facility, or if the death did not occur as the result of RRA, such as a premeditated event in a husband/wife pair, or if ambiguity was present. We also excluded events occurring in senior communities (i.e., residential areas that include single or multifamily residences and have age restrictions for residence). A total of 101 cases remained for further analysis. The first author located 22 supplemental media reports where available, by searching the Internet for pertinent details of each case.
Quantitative Analysis
Following identification of eligible cases, we created and assigned a new set of categorical indicators to capture trends through the cases. In some instances, we refined and added levels to available NVDRS categorical indicators (e.g., relationship between victim and exhibitor) and in other instances we created new indicators (e.g., creation of a single indicator to describe both victim and exhibitor sex in a given incident). As described above, we supplemented missing NVDRS data with data from available media reports, when it was clear based on date, location, and other details that the report described the same incident. We used R software (R Core Team, 2018) to identify frequency-based attributes of the sample using NVDRS-provided categories and our refined category list, and calculated means for age and age differences and time span between injury and death date.
Qualitative Analysis
We used a qualitative descriptive approach (Sandelowski, 2000) to analyze the narratives and supplementary media reports for data related to circumstances surrounding RRA events. Qualitative descriptive design does not utilize preselected variables (Sandelowski, 2000) and is therefore a naturalistic method of inquiry well suited for generating “answers to questions of special relevance to practitioners and policy makers” (p. 337).
Each of the three authors was randomly assigned a portion of the extracted narratives to code independently. We began our analysis with open first-cycle coding guided by Saldaña’s (2013) description of Initial Coding. According to Charmaz (2006), during Initial Coding the researcher should “remain open to exploring whatever theoretical possibilities we can discern in the data” (p. 47). Utilizing Chenail’s (2012) guidance, we identified qualitatively meaningful units of text for coding. We highlighted each meaning unit and recorded the code, line, and incident ID number in the comment bubble with features of Microsoft Word. Following first-cycle coding, we copied and pasted our comments to a new document where we began second-cycle coding following Pattern Coding to identify emergent themes by grouping codes into more meaningful units of analysis (Saldaña, 2013). In this phase, we clustered conceptually similar codes through a process of reflection and refinement, and individually developed a list of potential themes. We then reviewed the various themes and further refined these into four “super-ordinate” themes through abstraction, a basic process of identifying patterns between emergent themes, and clustering like themes, ultimately capturing a higher-level of connection between themes (Smith, Flowers, & Larkin, 2009). Finally, we each selected excerpts from the text, which we felt best represented the four themes, and through discussion of the essence of each theme selected the excerpts that we agreed support our findings.
Following we present findings from structured and unstructured data. For the sake of consistency, we use the terms victim and exhibitor in the findings. Additional information regarding the data analysis process is available by contacting the last author.
Findings
Structured Data
We identified a range of individual and incident characteristics. Most victims and exhibitors were White and male. Most frequent incidents involved two men, an exhibitor who was younger than the victim, occurred in a common area, were categorized as a “push,” did not involve use of a weapon (typically common objects), and involved two individuals with a dementia diagnosis. Average age difference between victim and exhibitor was 16 years (victims were older), and average time between injury and death was 17 days, with a range from 0 (death occurred on the same day as the injury) to 360 days. Most incidents (n = 51) took place in a skilled nursing facility. Another 20 took place in an assisted living facility; the remaining 30 occurred in facilities described using alternate terms (i.e., “home for elderly,” etc.). See Table 1 for victim and exhibitor characteristics and Table 2 for detailed case attributes.
Victim and Aggressor Characteristics From 101 Resident-to-Resident Aggression Cases.
Mental illness is derived from National Violent Death Reporting System narratives and includes both specific diagnoses (e.g., bipolar disorder) and generic descriptions of mental illness.
Attributes of Resident-to-Resident Aggression Cases.
Unstructured Data
Our four themes are unexpected, control, preventable, and absolution. Following we describe and define each and present supportive examples from the data.
Unexpected
This theme encompasses incidents which individuals described in narratives as occurring unexpectedly as well as those described as unexpectedly severe in nature of injuries. Most residents in our observations were reported to have multiple, chronic, and serious health diagnoses which did not necessarily predict imminent death but that could be better managed under specialized care in an LTC facility. Given the expectation that these residents would live out the remainder of their lives with elevated care and with other frail residents, many families and facility staff were blindsided by the occurrence of incidents that ended in a resident’s premature death. In several narratives, staff identified the exhibitor and victim as having been friends.
Prosecutors say [AGRESSOR] killed [VICTIM], . . . after the two women had an argument . . . the facility twice presented [VICTIM] with the chance to change either rooms or roommates, but she declined each time. He compared the pair to “sisters,” saying they took “daily walks together . . . ate lunch together every day, and were heard at night saying, ‘Good night, I love you,’ to each other.”
Many of the confrontations between victim and exhibitor consisted of a push or a slap resulting in the victim’s fall which caused further injury. However, the seriousness of such injuries was not always apparent; therefore, insufficient triage of the victim sometimes led to absence of timely medical intervention and consequently, an unexpected death.
Victim was . . . assaulted by another patient . . . Nursing home staff checked Victim’s injuries and did not send him to the hospital immediately. The next day, nursing home staff noticed that Victim’s face was swelling. Victim was transported to the hospital where he was admitted and pronounced 5 days later.
In contrast to serious injuries that were not immediately detectable, we also observed cases that were unexpectedly severe. It seems counterintuitive that an older, frail individual would have the bodily strength to inflict fatal harm on another, although due to these individuals’ frailty, little force is required to cause injury. The following describes a daughter’s reaction to encountering her mother following an incident of RRA: [VICTIM’S DAUGHTER] was the first to walk into her mother’s room. “They opened the curtain, and I screamed,” she said. . . The graphic photos [VICTIM’S DAUGHTER] took showed her mother black, blue and bloody. Her neck, nose and jaw were broken. So were 11 of her ribs and the bones on the right side of her face. Her lung had collapsed and had to be re-inflated twice.
Control
This theme includes incidents involving individuals’ efforts to regain or assert control over assigned or claimed personal spaces, as well as the use of common spaces.
[EXHIBITOR] killed [VICTIM] . . . after the two women had an argument over a table [EXHIBITOR] had placed at the foot of [VICTIM]’s bed. . . . [VICTIM] complained that the table obstructed her path to the bathroom, authorities said.
Frequently, incidents occurred following one individual entering a bedroom. In some cases, it was the exhibitor who entered, but in many, the victim either intentionally or accidentally entered the room and the exhibitor responded physically.
The Victim followed the [EXHIBITOR] into the [EXHIBITOR]’s room and told the Victim to get out of his room. When the Victim did not leave, the [EXHIBITOR] pushed the Victim who fell to the floor.
In addition to attempts to assert control over their assigned spaces, several incidents occurred when a resident became agitated over the use of common areas.
[EXHIBITOR] attempted to grab Victim’s walker, and Victim slapped [EXHIBITOR]’s hand in return . . . this altercation was believed to occur because [EXHIBITOR] wanted to walk the same path as Victim.
Preventable
This theme relates to incidents where an altercation between the victim and exhibitor escalated over a period of time during which intervention may have prevented injury or death, as well as incidents perpetrated by an individual who had demonstrated a pattern of aggressive behavior in the past.
Most individuals in our observations were diagnosed with dementia. Despite the tendency of dementia patients to become confused, agitated, and even aggressive, many incidents occurred in the absence of staff. One such incident reportedly took place over a lengthened time period, as the suspect made visits to the front desk to request water after which he returned to the victim to continue the assault: The [EXHIBITOR] stopped his attack twice to get a drink of water . . . after killing the victim, the [EXHIBITOR] walked down to the nurses’ station, requested another drink of water and told a nurse that he had just killed the victim.
Incidents also occurred in the presence of staff, who were hesitant or unable to intervene, or when the intervention strategies that were employed were insufficient.
[VICTIM] and [EXHIBITOR] had a [HISTORY] of combative behavior during which they would be physically separated and redirected to de-escalate the issues. On the morning in question one of these combative incidents occurred during which [VICTIM] confronted [EXHIBITOR] and [EXHIBITOR] picked up a vase that was sitting on a table. The two were standing in a face-off tossing threats of physical harm at each other. The caregiver/witness to this incident did not want to physically get between them because she was pregnant, but she attempted to unsuccessfully talk the men down. [EXHIBITOR] struck [VICTIM] on the head with the vase and pushed him.
As stated, many residents had a recent or ongoing history of aggression, both toward staff, other residents, and often, the victim.
Two months before the victim’s death, the [EXHIBITOR] is reported to have physically attacked him . . . an administrator “advised the two of them to get along” after learning of the alleged attack. The victim asked to be moved to a different room away from the [EXHIBITOR] but his request was denied. . . The homicide is believed to have arisen from another disagreement between the victim and [EXHIBITOR].
We observed many events that started with a verbal altercation and escalated into a physical confrontation, seemingly allowing opportunity for staff intervention.
Reportedly [VICTIM] had gotten into a verbal combative dispute with his roommate at the facility when [EXHIBITOR] struck [VICTIM] in the head.
Absolution
Within the absolution theme, we identified two subthemes that serve as defenses in absolving any blame otherwise placed on the exhibitor: cognitive decline and no intent to cause harm.
Cognitive decline
This subtheme encompasses the role of cognitive impairment in the exhibitor’s actions. As previously mentioned, in most of the narratives, at least one party, if not both victim and exhibitor, are described as having a dementia diagnosis. It was often suggested within narratives that confrontations among dementia patients were commonplace and uncontrollable. In the following excerpt, due to cognitive decline, the exhibitor was not aware of his actions even moments after the event: Both the [VICTIM] and the [EXHIBITOR] have impaired cognitive status due to dementia. The administrator at the facility talked with the [EXHIBITOR] and she stated that he does not remember anything.
In many instances (n = 33) it was explicitly stated that criminal charges would not be filed on the basis of the exhibitor’s mental status.
. . . the 77-year old resident who caused the death is sadly caught in the throes of severe dementia . . . he was unable to converse cogently with the detective about the incident and seemed unable to understand who the detective was or what he was investigating. This severe cognitive disability was confirmed by staff of the facility and available documentation. “In light of these factors, I find that the public interest is not served by seeking an indictment.”
No intent to cause harm
This subtheme represents the unintentional nature of the outcomes associated with the exhibitor’s actions.
The [VICTIM] told the [EXHIBITOR] to leave her alone and she pushed his hand away. Then the [EXHIBITOR] pushed the [VICTIM] and she fell striking her head on the floor. The [EXHIBITOR] tried to help the [VICTIM] up and said that he was sorry.
The following excerpt describes, from a legal viewpoint, the difference between an intentional action and an intent to bring harm and why the exhibitor was absolved of blame: Manner of death is homicide and was assigned solely because this was an intentional action by another patient in the medical facility where they were both patients. It does not imply criminal intent to cause harm.
The final excerpt illustrates absolving blame with the inclusion of both subthemes, cognitive decline and no intent to cause harm: The administrator of the nursing home stated about the [EXHIBITOR], “She is suffering from late-stage Alzheimer’s with little or no memory recall. I do not believe that she had any intent to harm the other resident, the [VICTIM], and she had no recollection of the incident within minutes after it occurred.”
Discussion
The purpose of this report was to analyze structured and unstructured data reflecting 101 incidents of fatal RRA identified in the 2003-2016 NVDRS data, to compare our results with prior research, to develop recommendations for future research, and to offer insights which may be useful for RRA prevention.
The results of our descriptive analysis of revealed multiple trends that are consistent with findings from prior research conducted in the United States and elsewhere. The results of our in depth qualitative analysis provided some nuanced insights into individual-level perceptions and behavior patterns that both confirm and enhance prior research results.
Consistent with both previous reports on RRA-related fatalities (Caspi, 2018; Murphy, Hons, Bugeja, Pilgrim, & Ibrahim, 2017), we identified “push-fall” incidents as the most common incident type, and head injuries, followed by hip fracture, as the most common injury location. Other consistencies with prior research include higher proportion of male exhibitors, reported by Caspi (2018), Gimm et al. (2018), and Murphy et al., and a meaningful age difference between exhibitors and victims, with victims typically older. As noted above, our analysis revealed a mean difference of 16 years between victims and exhibitors, while Caspi identified that exhibitors were on average 9.3 years younger than victims. Murphy et al. reported an average difference of 6.1 years. The combination of push-type incidents, male gender, and younger aged exhibitors all suggest that physical strength is a factor in RRA and is very likely a contributing factor to fatal RRA.
Another consistency between our findings and previous research is the prevalence of dementia diagnosis. In 70 of 101 cases we assessed, or roughly 70%, either or both parties had a dementia diagnosis. Murphy et al. (2017) reported that in nearly 90% of fatal RRA events, one or both parties had a diagnosis of dementia.
Among prior researchers, only Caspi (2018) explicitly described the time span between injury and death date. Caspi reported that the majority of victims (75%) suffered injuries that resulted in a decline of their physical condition over time rather than in an immediate fatality. We found a mean difference of 17 days between injury and death data with only 11 residents of 101 dying on the same day the injury occurred.
Our qualitative findings revealed similar themes to those presented by some researchers (e.g., Lachs et al., 2007; Pillemer et al., 2011; Snellgrove et al., 2013) including the suggestion that incidents might be unexpected, both in terms of the incident itself and the outcomes, and that incidents might be demonstrations of desire for control that began or became violent. Gimm et al. (2018) concluded residents with dementia have a fivefold increase in the likelihood of engaging in physical abuse; in support of this, we found evidence from multiple cases that exhibitors were deemed to lack cognitive capacity to be held accountable for their actions.
Our theme preventable might appear in conflict with our descriptive finding of no explicit triggering incident in 47 of 101 cases. A partial explanation is found in the assertion by Murphy et al. (2017) that the majority of exhibitors in fatal RRA events have a history of behavioral problems. Along with this, Ferrah et al. (2015) and Shinoda-Tagawa et al. (2004) suggested victims might also have a history of demonstrating disruptive and inappropriate behaviors. It is possible, therefore, that triggers are found in previous or ongoing occurrences and not just events proximate in time to a given RRA event. This suggests that although staff might be aware of potential for violence among some LTC residents and steps should be taken to mitigate that potential, victim behaviors that appear at first glance less harmful (e.g., wandering throughout common areas of a facility) might also be an appropriate intervention focus.
McDonald and colleagues (2015) reported three priorities for future RRA research: (a) developing and assessing interventions, (b) identifying environmental triggers, and (c) RRA incidence and prevalence. However, prior researchers (Ferrah et al., 2015; Rosen et al., 2016) have suggested that RRA events are infrequently documented or reported, so an important precursor to addressing these priorities is to improve identification and reporting of RRA events.
Although LTC facilities do not necessarily habitually train staff on recognition or prevention of RRA, promising results from Teresi et al. (2013) suggested substantial improvements in rate of reporting RRA events following implementation of a staff training program. We suggest emphasis on improved consistency and frequency of documentation of RRA is a critical step toward allowing LTC staff to identify patterns of behavior and to take preventive measures where appropriate. The delay we identified between injury and death date suggests that immediate injuries may appear less serious, so it is plausible that fatal RRA events may occur more commonly than is presently known. Improvements in event reporting would thus help more accurately capture overall U.S. trends in frequency and severity of RRA events. We provide these recommendations with awareness that in many facilities limited resources represents profound challenge in adapting procedures or increasing staff responsibilities.
In addition to recognition, prevention strategies should consider our findings in the context of organization-level policies and procedures. Most notably, roommate assignments and the need for reassignments should be made to minimize conflict. Some authors have gone so far as to suggest that LTC facilities should primarily offer single rooms to reduce RRA incidents (e.g., Caspi, 2018), and although we recognize the practical limitations to this recommendation, we emphasize that shared bedrooms represent a risk for RRA in both the aspect of relinquished control as well as the potential for unwitnessed events. In addition, this and other research has identified large age gaps as a potential risk factor for fatal RRA events. Although we are not able to estimate to what extent these age gaps play a role in triggering the events, we suggest the tendency for exhibitors to be substantially younger likely plays a role in the severity of the injuries sustained.
We make two recommendations for future research. First, to the extent possible, we suggest researchers assessing RRA incidents should also consider long-term follow-up. Two reviews indicated that there have been no studies on the long-term outcomes of RRA (McDonald et al., 2015; Rosen et al., 2008). Second, we recommend researchers place more emphasis on RRA in assisted living facilities. Our study is one of two that assessed a variety of LTC facility types (see Caspi, 2018), while the remaining body of literature focused solely on skilled nursing facilities. Our findings suggest that while the bulk of these events involve patients who require skilled nursing due to cognitive decline or physically ailments, RRA fatalities can and do occur in other types of facilities.
The primary limitation of this study is that, due to the nature of our data, these events we analyzed for this report only represent RRA events that resulted in fatalities that were reported to law enforcement agencies as such. This also means we were not able to capture fatal RRA that was not recognized or reported, or RRA that was not physical in nature, other than to the extent details of verbal or other aggression were captured in narratives of fatal events. In addition, although the combination of law enforcement and medical examiner reports contributed to comprehensive records, there were also missing demographic and diagnostic data, generally regarding exhibitors, and, in many instances, it was not straightforward to identify type of facility. Given important differences in how skilled nursing and assisted living facilities are staffed, regulated, and financed, the ability to associate incidents with facility types adds important insight. That said, although it is possible that examination of nonfatal RRA incidents might reveal different or additional trends, we believe the generally detailed and consistent nature of these data provide an important contribution to understanding trends in RRA incidents. In addition, our findings from structured and unstructured data reflected similarity to findings from prior research reports that included nonfatal incidents (e.g., Gimm et al., 2018; Snellgrove et al., 2013), which suggests potential similarity of trends in the incidents themselves regardless of the severity of the final outcome. Given eventual inclusion of reports from all 50 states in the NVDRS dataset, we suggest ongoing efforts to conduct in depth analyses of the structured and unstructured segments of these data has potential to continue to provide critical insights into RRA.
Footnotes
Authors’ Note
The National Violent Death Reporting System (NVDRS) is administered by the Centers for Disease Control and Prevention (CDC) by participating NVDRS states. The findings and conclusions of this study are those of the authors alone and do not necessarily represent the official position of the CDC or of participating NVDRS states.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
