Staff Reports of Behavioral Expressions of Persons With Dementia in 250 Assisted Living Communities

Abstract

Background and Objectives

Assisted living (AL) residents with dementia commonly exhibit behavioral expressions (BEs), yet no study has examined how AL staff perceive and respond to BEs in terms of the “ABC” model of antecedents, behaviors, and consequences, or how perceptions relate to organizational characteristics. Understanding staff perceptions may inform interventions.

Research Design and Methods

A convergent, mixed methods design was used in a study of health care supervisors from 250 AL communities in 7 states who reported 366 cases of resident BEs (one successful and one unsuccessful case). Qualitative analysis identified antecedents, BEs, staff responses, resident outcomes, and disposition (aging in place or discharge). Content analysis identified themes and compared case types. Descriptive statistics examined organizational characteristics associated with identifying antecedents.

Results

One quarter of cases recognized antecedents; slightly more were identified in successful (28%) compared to unsuccessful cases (20%); staff in dementia-only and smaller communities identified antecedents more often. Combativeness and anxiety were the most frequently reported BEs. The majority of both types of cases reported staff responses. Medication management was enacted as a response in 40% of cases compared to psychiatric assessment in 33% of cases.

Discussion and Implications

Staff training is indicated to increase AL staff recognition of antecedents; doing so might reduce the use of antipsychotic medications. Psychiatric assessment plays an important role in dementia care in AL and warrants further examination. Results could be helpful for applied behavioral researchers interested in developing ways to improve the identification of antecedents of BEs of persons with dementia.

An estimated 42% of the more than 811,000 assisted living (AL) residents in the United States have dementia (National Center for Health Statistics, 2018; Zimmerman et al., 2014). The majority of individuals living with dementia express behaviors and experience psychological symptoms that indicate a mismatch between their ability to cope and the demands in the social and physical environment which tend to worsen with the progression of the disease (Algase et al., 2007; Caspi, 2013; Cunningham, 2006; Mahoney et al., 2000; Teri & Gallagher-Thompson, 1991; Volicer & Hurley, 2003). Common behavioral expressions include agitation, depressive symptoms, aggression, and sleep disturbances, among others (Chen et al., 2000; Ismail et al., 2016; Volicer, 2020). These behaviors are also stressful to family members of persons with dementia, as well as to their care providers (Gilhooly et al., 2016; Stall et al., 2019). Strategies to address these behaviors include the use of psychoactive medications (Kales et al., 2019; McDermott & Gruenewald, 2019; Rattinger et al., 2013), nonpharmacologic approaches (Scales et al., 2018), and caregiver training (Hobday et al., 2010; Karlin et al., 2014; Pleasant et al., 2017; Teri et al., 2005). While many states’ AL regulations permit or require AL communities to limit the admission and retention of individuals with behavioral expressions (Carder et al., 2016), for residents with dementia, an average stay of 22 months (National Center for Health Statistics, 2018) obligates care staff to effectively prevent and manage residents’ behavioral expressions within the AL setting.

A prominent model used to educate dementia caregivers to address behavioral expressions that have a negative effect on the person living with dementia or others is the Antecedent–Behavior–Consequence (ABC) model (Beck et al., 2002; Teri, 1994; Teri et al., 1997; Teri & Gallagher-Thompson, 1991; Volicer, 2018; Volicer & Hurley, 2003). A long history in applied behavioral analysis incorporates training caregivers of persons with dementia in the ABC approach in multiple settings. Using this model, caregivers are trained to identify antecedents of behaviors and consider the consequences (immediate responses) that may reinforce that behavior, with the ultimate goal of preventing or deescalating behaviors (Burgio & Burgio, 1986; Karlin et al., 2014, 2017; Kong, 2005; Smith & Buckwalter; 1993; Teri & Gallagher-Thompson, 1991; Teri et al., 1997; Volicer & Hurley, 2003). Some early work includes Burgio and Burgio’s (1986) use of the ABC approach in Resources for Enhancing Alzheimer’s Caregiver Health interventions in community-based settings. Teri et al. (2005) implemented one of the first staff intervention studies (STAR—Staff in Assisted Living Residences) in AL using the ABC approach that resulted in significantly reduced levels of affective and behavioral distress.

However, the ABC model has not been used to examine how AL staff conceive of and respond to residents’ behavioral expressions, and its application may shed light differentiating successful from unsuccessful care and suggest areas for future intervention. Therefore, this qualitative study employed the ABC model to examine (a) what resident behavioral expressions staff report as most disruptive, and what they perceive to be the antecedents of those behaviors; (b) what differentiates situations seen as successful from situations that were not successful; (c) staff responses enacted to respond to behavioral expressions, and resident outcomes and related disposition (aging in place or discharge); and (d) whether certain types of AL communities are better able to identify antecedents of behaviors. The latter question was based on a hypothesis that antecedents would more likely be recognized in AL communities that were smaller, dementia-specific, and had nurses on staff. Overall, it was of interest to better understand situations in which behavioral expressions might be anticipated and potentially prevented.

Method

This study employed a convergent mixed methods research design (Creswell & Clark, 2018). A cross-sectional interview of AL staff included an open-ended question that provided narrative data. Content analysis, both qualitative and quantitative, was used to analyze these data (Cho & Lee, 2014). This approach concurrently integrates both qualitative and quantitative sources of data to draw on strengths and weaknesses of quantitative (close-ended) and qualitative (narrative responses) data for a more complete understanding of the study topic, and how AL staff understand and respond to residents’ behavioral expressions.

Sample

AL communities from seven states participated in this study—Arkansas, Louisiana, New Jersey, New York, Oklahoma, Pennsylvania, and Texas—chosen because they are located in census divisions that reported a range of antipsychotic prescribing practices which was also a topic of study. To facilitate on-site data collection, two geographic regions were constructed in each state to represent the entire state in regard to eight demographic variables used in other work: per capita income, percent of the population below the poverty level, percent of population non-White, unemployment rate, percent of the population aged 65 and older; number of primary care physicians, and hospital and nursing home beds per individual aged 65 and older (Zimmerman et al., 2005). In AR, LA, NJ, and OK, the regions encompassed the entire state. Eligible AL communities were those that provided nonnursing, residential, long-term care to more than four adults older than 65 years of age (N = 1,624). With the intent to recruit 35–40 sites per state (total N = 250), communities were randomly sampled proportionate to the size and invited to participate. Participating AL communities received a $100 gift card.

Data Collection

To reduce possible historic effects across states, data collection occurred in one half of the communities in each state and then in the second half (October 2016–November 2018). Data used in this article are those obtained in interviews with the administrators and the “health care supervisor” at each site (i.e., the individual responsible for health care provision and most knowledgeable about residents’ health); interviewees included licensed nurses, certified nurse aides, administrators, wellness directors, and resident care managers. For the purpose of this article, we use the term health care supervisor to represent interviewees, all of whom provided informed consent to participate; study procedures were approved by the Office of Human Research Ethics of the University of North Carolina Chapel Hill.

Measures

The primary focus of this article is health care supervisors’ responses to an overall question with two prompts: Thinking about residents with dementia who have behavioral symptoms such as agitation, aggression, resistance, wandering, or other behaviors, or problems with sleep or mood … (1) Think about some of the most disruptive residents who’ve lived here. Can you tell me about one of those residents whom you see as a real success—meaning, a resident with serious behavioral symptoms who was helped through successful management by you or the other staff? This should be someone about whose care you feel proud or satisfied. Can you tell me that story? (2) Now can you tell me about one of your residents whose behavioral symptoms were so severe that you or your staff weren’t able to manage them? Can you tell me that story? (See Supplementary Figure 1 for related prompts.) In addition, data were obtained regarding characteristics of the AL community, including the presence of a dementia-specific care area (memory care), bed size, and presence of a nurse.

Analysis

Content analysis (Hsieh & Shannon, 2005) was used to understand the narrative responses to the two prompts. Specifically, we used both a directed approach, in which analysis starts with a theory or framework to guide initial code development, and summative content analysis, which involves counting codes as well as interpreting the meaning of narrative texts. Two sets of codes were developed a priori by an interdisciplinary team of seven researchers trained in the fields of sociology, gerontology, social work, nursing, and health services. We used rigorous methods to assure data validity, including team coding and interrater agreement. All qualitative coding and analysis were conducted using Atlas.ti version 8 (Scientific Software Development, 2018).

The first set of codes, developed using the directed approach to content analysis, was informed by the ABC model described above. Specifically, that theory-based conceptual model relates to antecedents (something that triggers a behavior and is typically potentially modifiable), resident behaviors (labeled behavioral expressions), and consequences (behavior occurring immediately after the target behavior; Cunningham, 2006; Logsdon et al., 1998; Teri, 1994; Teri & Gallagher-Thompson, 1991; Volicer & Hurley, 2003). In “telling the story,” staff’s narratives described their perceptions of antecedents, behavioral expressions, and their own responses to resident behaviors. In some cases, the described responses were not always those that occurred immediately after the behavior (e.g., medication management, psychiatric assessment) and so did not constitute “consequences” in the pure sense of the ABC model. Our adapted version of the ABC model therefore included antecedents, behavioral expressions, staff responses, and also resident outcomes (positive, negative, or general) and related disposition (aging in place or discharge from AL; Supplementary Figure 2). This approach, also called deductive category application, provides a way for qualitative researchers to use a theoretical framework or conceptual model to develop a coding scheme and understand relationships between codes (Hsieh & Shannon, 2005).

After reviewing and discussing first-level codes, the second set of descriptive codes was developed for all categories to identify common concepts and keywords used in the narratives. This emergent approach is similar to grounded theory, as it is both emergent and based on interviewees’ words. In some cases, subcategories were developed when codes grouped together.

To assess the reliability of the code sets, two 2-person teams coded 70 responses separately, then met regularly with the entire team to reach a consensus about codes reflected in each response. This iterative process was employed to establish a coding protocol (based on 55 resident cases) and test interrater agreement (based on 15 of each case type, successful and unsuccessful). Once interrater agreement consistently exceeded 80%, the data were divided so that each coding team was assigned one half of the total cases, which included 213 responses to the first question about successful cases and 153 responses to the second question about unsuccessful cases (N = 366 total cases). All seven members of the investigative team provided input on discrepant codes until a clear and agreed-upon description of codes resulted.

Percentages of all cases, and separately for successful and unsuccessful cases, were computed to determine the frequencies of antecedents, behavioral expressions, staff responses, resident outcomes, and dispositions. An analysis of co-occurrences in Atlas.ti of multiple strategies was conducted to compare successful and unsuccessful cases. Quotes are used to illustrate the meanings and context of these cases.

Results

A total of 743 AL communities were invited to participate; of these, 354 (48%) administrators refused; the recruitment of 130 (17%) remained pending at the conclusion of the study; and data collection in nine (1%) was not completed. Nonparticipating and participating sites did not differ by size (p = .43). Table 1 summarizes the characteristics of the AL sample. The majority of communities were for profit (71%) and operated in association with either another AL community, nursing home, or continuing care retirement community (55%). The average number of years in operation was 23, and the average bed size was 54; the majority did not have any dedicated dementia care beds (54%), and a minority were dementia-specific (10%). More than 60% of the residents were 85 years of age or older (61%); the majority were female (72%) and White (96%), and on average, 41% had a diagnosis of dementia.

Table 1.

Assisted Living Characteristics (N = 250)

Variables	N (%) or M (SD)
Organizational characteristics
State
Arkansas	35 (14.0)
Louisiana	27 (10.8)
New Jersey	37 (14.8)
New York	38 (15.2)
Oklahoma	37 (14.8)
Pennsylvania	40 (16.0)
Texas	36 (14.4)
Urbanicity
Metro: 1,000,000+	128 (51.2)
Metro: <1,000,000	88 (35.2)
Nonmetro	34 (13.6)
Ownership
Profit	176 (70.7)
Nonprofit	73 (29.3)
Affiliated with a religious organization
Yes	27 (11.1)
No	217 (88.9)
Years in operation	22.8 (25.9)
Bed size	54.4 (31.4)
Dementia care beds
None	135 (54.0)
Some	91 (36.4)
All (entire community is dementia care)	24 (9.6)
Dementia bed size (proportion of bed size)	20.0 (30.2)
Case-mix
Age (%)a
<65	4.8 (11.3)
65–74	9.2 (9.3)
75–84	25.4 (9.5)
85–94	49.7 (16.0)
≥95	10.9 (6.8)
Gender, female (%)a	71.9 (11.8)
Race (%)a
White	95.9 (9.7)
Black	3.5 (9.4)
Other	0.6 (1.9)
Ethnicity, Hispanic origin (%)a	3.9 (13.4)
Cognition, function, mental illness (%)a
Diagnosis of dementiaa	40.9 (26.1)
Short-term memory problems or frequent disorientation	21.7 (25.6)
Require staff attention for behaviors	19.0 (23.8)
Incontinent of urine	47.0 (27.5)
Regularly use a wheelchair	15.7 (16.8)
Intellectual/developmental disability	1.4 (4.0)
Mental health diagnosis	33.3 (17.5)
On hospice (%)a	6.6 (9.7)
Receiving Medicaid (%)a	10.1 (24.3)

^aCalculated based on the weighted number of residents/site.

The majority of health care supervisors were female (95%), White (65%), and non-Hispanic (80%). Close to one third were registered nurses (RNs; 32%) and almost half were licensed practical nurses/vocational nurses (46%). Other respondents identified themselves as administrators (11%) or care aides (3%) who were serving as the health care supervisor.

Components of the Expanded ABC Model

The 250 respondents provided responses for 213 “successful” cases and 153 “unsuccessful” cases for a total of 366 cases; not all respondents provided both a successful and unsuccessful case example.

Antecedents

The coding guide described an antecedent as something that triggers a behavioral expression and is typically potentially modifiable (an example being a staff report of a family having moved a picture in the resident’s room as triggering behaviors). It can include something that is in the mind of a person with dementia (e.g., thinking someone stole something from them). To be clear, it cannot be certain that the named antecedent was actually causal, but it was perceived by staff in that way. Of all reported cases, 25% had a reported antecedent (28% of successful cases and 20% of unsuccessful cases, indicating that there were instances when antecedents were identified but the behavior was not be mitigated). Supplementary Table 1 provides examples of perceived antecedents. We also examined differences in reporting an antecedent based on organizational characteristics (Supplementary Table 2). Although the fewest overall cases were reported by respondents who worked in dementia care only areas (in part because they represent a minority of the entire sample), they identified proportionately more antecedents among those cases than staff who worked in other areas (30% compared to 23%–25%). Similarly, staff who worked in smaller communities identified antecedents for 37% of cases, compared to 19%–29% in larger communities. There was little difference in the identification of antecedents based upon whether the community had nurse staffing (25% compared to 27%). In all comparisons except one, proportionately more antecedents were identified in successful compared to unsuccessful cases. The largest difference was for AL communities that had only dementia beds; in these communities, antecedents were identified in 42% of successful cases compared to 13% of unsuccessful cases.

Behavioral expressions

Our analysis identified 11 types of behavioral expressions that included (in alphabetical order) abuse, anger, anxiety, cognition, combativeness, elopement, resistance to care, restlessness, socially inappropriate behavior (SIB), suicide, and verbal abuse. Table 2 provides examples of successful and unsuccessful exemplar cases of each behavioral expression; some of these are not complete sentences because they are verbatim notes that data collectors wrote as the respondent talked (i.e., noncritical words were omitted on occasion). The two most common behavioral expressions reported by staff were combativeness (29% all cases; 26% successful, 32% unsuccessful) and anxiety (28% all cases; 32% successful, 24% unsuccessful). There were three types of combativeness: physical aggression toward staff (struck out at a staff member), physical aggression toward another resident (struck out at a resident), and physical aggression but not toward another person (e.g., struck out at an object), the latter of which was the most common type. Medication management was a typical staff response for combativeness. For example, “She was very disruptive and physically aggressive. She got meds for behavior. She later stated that she doesn’t remember doing any of that. A referral was made to [doctor] and the resident got on new meds.” Compared to combativeness, anxiety was more frequently reported as resolved successfully than unsuccessfully. The following is an example of when a staff response was successful: “A resident experienced extreme anxiety, crying hysterically, we realized it’s because of pain. She has a pump in back that releases pain meds. We followed up with doctor to increase medication.”

Table 2.

Percent of Times Components of the Expanded ABC Model Were Reported and Quotation Example by Type of Case (N = 366)

Component of ABC model	All cases (N = 366)	Successful cases (n = 213)	Unsuccessful cases (n = 153)	Quotation example
Antecedent (N = 90): Something that triggers a behavioral expression and is typically potentially modifiable	25%	28%	20%	Successful case: “Family moved picture in room which triggered behaviors.” Unsuccessful case: “When it’s later in the evening (sundowning).”
Behavioral expressions (N = 425 identified by staff): Refers to what the resident did/the behavior that was described; may be a single behavior, several different behaviors, or a recurring behavior
Abuse (n = 20): Any mention of abusive behavior	6%	2%	10%	Successful case: “One male resident who would curse and fight (punching, biting). Would happen (during bath). Sent resident to behavioral psych; returned after one month. No longer fights staff. Less memory problems and behaviors due to new meds.” Unsuccessful case: “Nurse said he has behavior issues. … Only here a week but had outburst. When staff tried to wake him up he hit, scratched, punched and chased staff. First night here, he was very rude, used foul and racist language.”
Anger (n = 39): Mention of aggression, aggressive behavior, or anger	11%	13%	6%	Successful case: “Resident with sudden anger outbursts who is redirected easily when spoken to. Resident takes Sertraline.” Unsuccessful case: “[Resident] would not let anyone change or wash her. Would sit in urine … would get angry. Got psych consults, palliative care for her but not successful.”
Anxiety (n = 104): Anxiety, agitation, worry; includes screaming and paranoia (only coded when it was the behavior, not the antecedent)	28%	32%	24%	Successful case: “Resident that gets really agitated on a daily basis. Does not recognize place so hard to redirect. Found it helpful to go with it ... not tell her she has been here for years, but help her with where she is at.” Unsuccessful case: “Symptoms included severe agitation, depression, abdominal pain. … One night staff went into [room] and she tried to hurt herself with steak knife. Went to behavioral hospital.”
Cognition (n = 50): Mention of specific cognitive expressions; examples include “confused,” “disoriented,” “decreased attention span”	14%	18%	8%	Successful case: “Resident confused, disoriented, sundowning. Primary care physician, psych and education have helped. Coloring, pancakes, targeted enjoyment of activities, interaction with staff; Ativan PRN manages pain with fentanyl patch, Seroquel.” Unsuccessful case: “We had a wanderer, very confused, to the point he would start exit seeking. He urinated in cabinets. One night he was so violent even when he has been redirected. He would look into the mirror and not recognize himself. He needed care we couldn’t give.”
Combativeness (n = 105): Violent, resistant, and aggressive physical behavior	29%	26%	32%	Successful case: “She was very disruptive and aggressive. She got meds for behavior. She later stated that she doesn’t remember doing any of that. A referral was made to [doctor] and the resident got on new meds. After a week, she was on nursing duty helping others.” Unsuccessful case: “Had a male resident who was very young. Had alcohol induced dementia. Would urinate on plants. Very sexually aggressive and inappropriate with staff thinking they were bartenders. He became physically aggressive.”
Elopement (n = 43): Exit seeking; wandering itself is not elopement	12%	12%	11%	Successful case: “Used to volunteer here before. Declined rapidly and left building, walked to house and fractured arm and fell. Moved to dementia care unit. Calm and volunteers. Give her tasks to complete.” Unsuccessful case: “One woman would run faster than me and head to the highway.”
Resistance to care (n = 53): Refusing or resisting any type of care	14%	15%	14%	Successful case: “Resident never washes herself and one of the staff members was able to get her to shower on her own.” Unsuccessful case: “Gentleman with Parkinson’s, clinically complex, required extensive assistance with care, he was not compliant with his diet, went into renal failure and was hospitalized.”
Restlessness (n = 58): Restlessness, pacing, wandering	16%	18%	13%	Successful case: “Resident who had wife in independent living—he wanted to go with her, but he had Parkinson’s and dementia. We came up with a schedule for visiting; he wouldn’t wear a wander guard device so we put it on his ambulation devices—successful because it allowed him to roam freely when appropriate.” Unsuccessful case: “Wandered during the night, thinking parents were still alive. … Spiraled so much, daughter not accepting of challenges, had an acute episode.”
Socially inappropriate behavior (n = 37): Examples include public undressing or throwing feces, public urination; spitting on the floor; does not include cursing	10%	4%	17%	Successful case: “An elderly resident had aggressive behaviors, taking pictures off wall and throwing/breaking them. Agitated and combative, urinating and bowels in wrong places.” Unsuccessful case: “Someone had disruptive behavior who we had to let go. Undressing, sitting on public chairs and having bowel movements; getting into cupboard for food; spitting on the floor.”
Suicide (n = 8): Suicide ideation and any mention of an attempt to harm self and end one’s life	2%	1%	3%	Successful case: “Female resident, no family or friends. Had psychosis event in wanting to kill self.” Unsuccessful case: “A lady that just moved in, has multiple personality disorder, would talk about suicide every day and would steal scissors. She used fingernail polish as lipstick.”
Verbal abuse (n = 38): Verbally abusive; includes verbal threats	10%	11%	10%	Successful case: “Verbally abusive. Now one of the most active residents and helps take care of others. One-on-one and build trust with resident, worked with family to find out things she enjoyed.” Unsuccessful case: “She got more and more agitated and aggression started with husband then to staff. Went from verbal to physical. They both had to move out, husband and wife. Some med management even went to hospital on one depressant for anxiety (Buspar).”
Staff responses (strategies enacted or considered; N = 583): How staff reacted to the behavior; what was done to address the behavior
Health strategies (HS; n = 267 strategies in 215 cases)	59%	64%	51%	Successful case strategy considered (HS, Combative): “Talk of sending her away to psych hospital before sending to hospital. Put her on [an antipsychotic] and mood stabilizers.” Successful case strategy enacted (HS and SS, Socially Inappropriate Behaviors): “Heavily cued in meals and activities, constant reassurance. Took her off [psych meds] and put her on [blood pressure medication] and [listed medications].” Unsuccessful case strategy enacted (HS and SS, Verbal abuse): “… sent resident to doctor, talked to family. Redirection and medication didn’t work, and resident kept harassing woman she thought was trying to steal her man.” Successful case strategy enacted (SS, Verbal abuse): “Visit with him to figure out issue and redirect. Cannot communicate specific needs; sit with him and calm him down. Call daughters to relieve him. Must have newspaper before breakfast. Staff must be patient and issue/behavior resolves.” Unsuccessful case strategy enacted (SS, Abuse): “Resident struck another resident, multiple meetings with family, son was in denial.”
Medication management (n = 147)	40%	50%	26%
Psychiatric assessment (n = 113)	31%	28%	35%
Rehabilitation (n = 7)	1%	2%	<1%
Staff strategies (SS; n = 316 strategies in 217 cases)	59%	66%	50%
Activities (n = 47)	13%	20%	3%
Family involvement in care (n = 101)	28%	23%	34%
Person-centered care (n = 55)	15%	23%	4%
Police involved (n = 14)	4%	1%	7%
Redirection (n = 78)	21%	27%	13%
Validation (n = 21)	6%	9%	0%
Other (n = 29)				Unsuccessful case strategy enacted (Elopement): “Had to get her a 24-hour sitter until we could find her a new location.” Successful case strategy enacted (Combative): “Hospice resident was getting to point where he was getting aggressive, argumentative with staff. Hospice put him on meds. Then he was overmedicated. They backed him off.”
Home health (n = 15)	4%	4%	4%
Hospice (n = 14)	4%	4%	4%
Resident outcomes (N = 310 reported by staff): What the resident did/behavior described as having occurred—or not occurred—after an enacted staff response)
Positive (n = 198): Refers to the behavior that occurred/did not occur after an enacted staff response	54%	86%	9%	Successful case (Hospice strategy, Combative): “Now he is behaving great. No more behaviors, acting like himself.” Unsuccessful case (HS, Restlessness): “Still wanders halls and has manic episodes.” Successful case (HS, Suicide): “Medication has helped some, but still aggressive.”
Negative (n = 66): Refers to the behavior that occurred/did not occur after an enacted staff response	18%	4%	37%
General (n = 46): When the positive and negative nature of the account is unclear; mixed results	13%	14%	10%
Disposition (N = 109 reported by staff): Explicitly stated disposition for resident that indicates he/she is still living in AL or was discharged
Age in place (n = 14): The resident did not permanently leave AL; explicitly stated	4%	7%	0%	Successful case (SIB): “Fell, broke his hip, had to go to skilled, came back home and has gotten better, even stopped smoking.” Successful case (Combative): “Has been excellent … maintains same dementia diagnosis but much better.”
Discharge from AL (n = 95)	26%	4%	56%	Successful case (Combative): “Eventually found nursing home care placement.” Unsuccessful case (Anger): “Had to transfer to LTC where she will have more one-on-one care.” Successful case (Anxiety): “Now he is in dementia unit doing well.” Unsuccessful case (Restlessness): “Had to be discharged and admitted to a separate memory care community.”
Nursing home (N = 20): Explicitly stated that resident was discharged	5%	2%	10%
Memory care: Explicitly stated that resident was discharged to the memory care unit	9%	9%	8%

Notes: ABC = Antecedent–Behavior–Consequence; AL = assisted living; LTC = long-term care. Some quotes are not complete sentences because they are verbatim notes that data collectors wrote down as the respondent talked (i.e., noncritical words were omitted on occasion). Percentages for health and staff strategies represent a number of cases.

Restlessness was reported in 16% of all cases (18% successful and 13% unsuccessful). Wandering was a common example of restlessness, as illustrated in this unsuccessful case, “Resident who came from independent, had confusion, diagnosis of dementia, didn’t adjust well to transition, big wanderer, wandered out.” Resistance to care was reported in 14% of all cases (15% successful and 14% unsuccessful). Some of the common examples of resistance to care included refusal to bathe, get dressed, eat, and take medications. One example is when staff identified the antecedent for a woman who refused to eat at mealtimes, “Thought husband would be joining and she was waiting.” Another is staff describing how they worked with family and provided person-centered care to successfully manage resistance, “Would not change clothing for a week, verbally abusive. Now one of the most active residents and helps take care of others. One-on-one and built trust with resident, worked with family to find out things she enjoyed.”

The behavioral expressions more likely reported as unsuccessfully versus successfully resolved were abuse (10% unsuccessful and 2% successful), combativeness (32% unsuccessful and 26% successful), and socially inappropriate behavior (17% unsuccessful and 4% successful). Suicide was rarely reported (eight cases; 2% of total cases; 3% unsuccessful and 1% successful). In many instances, more than one behavioral expression a resident exhibited was unsuccessfully resolved, as in the following example of a resident who exhibited socially inappropriate and combative behavior, “Had a male resident who was very young. Had alcohol-induced dementia. Would urinate on plants. Very sexually aggressive and inappropriate with staff thinking they were bartenders. He became physically aggressive.”

Staff responses

Two overriding categories of staff responses emerged during coding: health strategies and staff strategies. A total of 267 health strategies were reported for 215 cases (59% of all cases). They included medication management (40% of all cases), psychiatric assessment (31%), and rehabilitation (1%). In 10% of all cases, it was explicitly stated that the physician was seen to adjust medications. These strategies were reported by staff in 64% of successful cases and 51% of unsuccessful cases. Psychiatric assessments included inpatient or outpatient visits and were enacted in 113 cases (28% successful and 35% unsuccessful). A successful case example is, “… psychosis developed, she went to a behavioral rehab for two weeks and came back a different person and better resident.” Sometimes psychiatric care did not resolve the situation and the resident was discharged permanently, “Upon admission, family said she likes to go outside. Did not disclose psychiatric background. Go out at night, walk around in wetlands, come back with bite marks. Sent her out to behavioral health few weeks stay. Was not effective. Resident put herself in dangerous positions. Had to tell family she needed to find another place.”

Medication management as a health strategy was enacted in 40% of all cases (N = 147) and twice as often in successful cases compared to unsuccessful cases (50% vs. 26%; Table 2). Staff reported 3 times more often increasing existing medications or adding a medication than decreasing a medication. The following is an example of a successful case where the staff increased medications in addition to providing person-centered care for elopement and restlessness behaviors, “A resident that tried to escape, we started her on Depakote two times a day, the behaviors weren’t changing. Added Seroquel one time a day, helped in the afternoon. She keeps searching for her dead babies so we placed a cradle for the baby outside her room with a doll so hopefully it will help.”

Other staff strategies (316 reported for 217 cases; 59% of all cases, 66% of successful cases and 50% of unsuccessful cases) included (in alphabetical order) activities, family involvement in care, person-centered care, police involvement, redirection, and validation. Family involvement in care was the most common staff strategy, followed by redirection and then person-centered care. These common strategies also involved activities, such as asking the family what type of activity the resident likes, redirection to do an activity, or individualized person-centered care involving the resident in a preferred activity. The following successful case example includes redirection, medication management, and activities, “Along with medication and redirecting and a lot of keeping them busy and providing activities. One resident was always trying to go home or go to work. Seroquel helped.” Successful cases enacted person-centered care, validation, and activities at a greater frequency than unsuccessful cases. Validation was not mentioned at all in unsuccessful cases; person-centered care was enacted in 4% of unsuccessful cases and activities in 3% of unsuccessful cases. Family involvement in care was enacted more frequently in unsuccessful cases (35%) compared to successful cases (23%), and family involvement in care for unsuccessful cases was more often negative. For example, a respondent described that one resident “had a lot of behaviors in reference to screaming and yelling … family was in denial about his condition and refused any interventions [extra home health, medications].”

In a small portion of cases (14), staff involved the police. The behaviors exhibited were either combativeness, abuse, or elopement for residents with psychiatric diagnoses. All but three were unsuccessful cases, and all unsuccessful cases resulted in discharge from AL. None of the 14 cases occurred in dementia-only AL communities or smaller communities (<24 beds). An analysis of co-occurrences of all strategies by case type indicated that multiple strategies were enacted in approximately 50% of successful cases compared to one third of unsuccessful cases.

Resident outcomes

The majority of the successful cases report a positive resident outcome (86%) compared to 9% of the unsuccessful cases; similarly, only 4% of successful cases reported a negative resident outcome compared to 37% of unsuccessful cases. Examples of positive and negative outcomes, respectively, are “Now he is behaving great. No more behaviors, acting like himself” and “he refuses to take his [medication] and is still very aggressive.”

Disposition

Very few cases reported a disposition of aging in place (N = 14, 4%)—which does not mean it did not occur more often, only that it was infrequently mentioned—all of which were reported as successful cases. Approximately one quarter (26%) of all cases reported discharge from AL, 4% of which were considered successful (because the person required more care than the setting could provide) and 56% unsuccessful. An example of a successful case where the disposition was discharge was the behavioral expression of elopement and the determination that the person needed a more secure unit, “Used to volunteer here before. Declined rapidly and left building, walked to house and fractured arm and fell. Moved to dementia care unit. Is calm and volunteers.”

Discussion

This study contributes to the literature on behavioral expressions and dementia care in AL based on 366 case reports provided by health care supervisors in 250 AL communities. The ABC model informed content analysis of a large sample of AL staff descriptions of behavioral expressions of persons with dementia, generating numerous novel findings: (a) staff infrequently conceptualize antecedents of residents’ behavioral expressions; (b) unsuccessful cases more frequently reported behavioral expressions of combativeness and socially inappropriate behavior; (c) in the majority (59%) of all cases, close to two thirds of successful and one half of unsuccessful cases, staff enacted health strategies (medication management and psychiatric assessment), and 59% enacted other strategies (e.g., activities, family involvement in care, person-centered care) as strategies to resolve behavioral expressions; (d) multiple strategies were enacted and considered (health and staff strategies) more often in successful cases compared to unsuccessful cases; and (e) cases perceived as successful overwhelmingly reported positive resident outcomes (86%) compared to just 9% of unsuccessful cases.

A critical finding is the infrequency with which staff conceptualized the antecedents of residents’ behavioral expressions: Only 25% of all cases reported an antecedent, and so even in the best of circumstances, health and other staff strategies were responsive to perceived antecedents a minority of times. Experts in dementia care have long stressed the importance of recognizing antecedents if behavioral expressions are to be avoided (Beck et al., 2002; Logsdon et al., 1998; Teri & Gallagher-Thompson, 1991; Teri et al., 2012; Volicer, 2009), but a recent article pointed out that some supervisors may take issue with time being spent on prevention rather than attending to daily care tasks (Fazio et al., 2020). We hypothesized that staff in dementia care units, smaller communities, and that had nurse staffing would be more likely to conceptualize antecedents; the first two hypotheses were borne out. All but six states in the country have training requirements for staff who work in dementia care units (Carder et al., 2016), which could potentially help staff better understand and identify antecedents. Examples of staff training requirements in some of the states in this study include eight additional hours specific to dementia care within the first 30 days (PA); 4 h prior to beginning the job, 16 h on-the-job training in the first 16 h, and 8 h annually (TX); and 30 additional hours per year of dementia-specific training (AR; Carder et al., 2016). Some states also require more licensed staff in dementia care areas including social workers (AR) and RNs and home health aides (PA, NY; Carder et al., 2016). These licensed professionals have additional training and knowledge related to person-centered care, which is essential to address behavioral expressions (Fazio et al., 2020). Few states require a social worker, however, despite the fact that they have been found to play a helpful role in AL (Zimmerman et al., 2013). Finally, the finding that staff who worked in smaller communities identified antecedents more often may be supported by the fact that staff-to-resident ratios are typically higher in smaller AL communities, which provides staff the opportunity to devote more time for person-centered care to residents, a key component to the conceptualization of antecedents.

Staff perceptions of antecedents to behavioral expressions are important to understand, especially for the most common unsuccessfully resolved behavioral expressions—combativeness and socially inappropriate behavior. Future interventions could help staff identify antecedents to these behaviors, perhaps using a true behavior analysis framework. AL communities could also consider implementing evidence-based dementia care training programs such as STAR (Teri et al., 2005) and CARES (Connect with the Person, Assess Behavior, Respond Appropriately, Evaluate What Works, and Share with Others) (Dobbs et al., 2018; Hobday et al., 2010; Pleasant et al., 2017) which focus on antecedents that trigger behaviors and staff immediate responses that are most appropriate to respond to those behaviors. In addition, implementation of care programs specifically designed to avoid behavioral expressions related to resistance to care and combativeness (e.g., Bathing Without a Battle, Mouth Care Without a Battle; Sloane et al., 2004; Zimmerman et al., 2020) would help identify antecedents to behavioral expressions. Also, family members are central to assisting staff in the identification of antecedents (Fazio et al., 2020). The findings of this study highlight the role of the family in responding to behavioral expressions; they need to work with staff for the best resident outcome.

Another important finding is that medication management and psychiatric assessment, including inpatient stays at a psychiatric hospital, were used as strategies in 40% and 31% of cases, respectively; a difference between the two, however, is that medication management was more often reported to result in a successful case, whereas psychiatric assessment was more often reported to result in an unsuccessful case. It is widely recognized that medications should be the last treatment option of choice (Zimmerman et al., 2015), and so their use, and the fact that they are considered effective, suggests more effort is needed to promote nonpharmacological practices to address behaviors. To staff’s credit, however, they enacted multiple strategies to address residents’ behavioral expressions in 50% of the successful cases and one third in the unsuccessful cases, indicating that medications and hospitalization were not the first or only treatment option. The role of psychiatric hospitals for AL residents with dementia has not been widely reported in the literature; they seem to perhaps be an avenue of last resort, and to the extent that they are sometimes considered effective, more research on the role and effectiveness of inpatient psychiatric hospital stays for AL residents is indicated. As with other health settings, psychiatric hospitals vary in their use of evidence-based dementia care practices, including the use of psychoactive medications (Goga et al., 2017).

Aging in place is a goal for many AL residents. Yet in some of the successful cases—and more often reported in unsuccessful cases—discharge from AL to nursing home or memory care occurred, presumably with the expectation that these settings were better equipped to deal with behavioral expressions. Our data support that memory care only AL communities are better at identifying antecedents to behavioral expressions, which more often result in successful cases. Discharges often occurred after employing other strategies, including referring the person to an inpatient psychiatric hospital and increasing medications. Because AL communities may not be able to provide sufficient care for all behavioral expressions, it is appropriate when they recognize that. A common reason given for discharge was “they needed more care than we could provide” or “they needed a higher level of care.” That said, there may be room for staff to change their perception of what is “unsuccessful,” by realizing that a more appropriate setting of care can be viewed as successful.

Limitations to this study include that the reports were retrospective and there could have been recall bias. Of course, the perspectives are from staff report only and may not be borne out in fact. That said, it is human nature to respond based on one’s perceptions, and so care provided by AL staff is grounded in their perceptions. In addition, staff were not specifically asked about their immediate response to resident behavioral expressions as defined by the ABC model. The fact that in many cases they instead discussed distal responses (i.e., medication management and psychiatric assessment) suggests that they do not typically think about their own immediate controlling response. Future research should explore more fully staff’s perceptions about the consequences (immediate responses) to resident behavioral expressions.

Conclusions and Implications

Findings from this study strongly support the need for AL staff to more often consider and respond to antecedents of the behaviors of persons with dementia; when reporting cases of behavioral expressions, they identified antecedents only 25% of the time and more often in cases that they considered to have a successful resolution. Focusing on antecedents may be a promising strategy to reduce the use of psychotropic medications, which were reported in 40% of cases. In addition, formal psychiatric assessment seems to play a notable role in dementia care in AL and warrants further study.

Funding

This work was supported by the National Institute on Aging of the National Institutes of Health (R01AG050602).

Conflict of Interest

None declared.