Delirium and Antipsychotic Medications at Hospital Intake: Screening to Decrease Likelihood of Aggression in Inpatient Settings Among Unknown Patients With Dementia

Abstract

For individuals with dementia, disorientation and both external and internal stimuli may trigger behaviors that are difficult to manage or dangerous to health-care providers. Identification of correlational risk factors to aggressive behavior in patients who are unknown to the hospital can allow providers to adapt patient care quickly. Records for patients aged 60+ who spent at least 24 hours at the hospital other than in the psychiatric unit were used (N = 14 080). The first 4000 records and every 10th person who met criteria (N = 5008) were searched for documentation of dementia (n = 505). Logistic regressions and χ² tests were used to examine relationships between variables. Recognition of delirium (P = .014, Exp(B) = 2.53), coupled with an existing prescription for antipsychotic medication at intake (P < .001, Exp(B) < 4.37), may be a reliable means of screening for risk and intervening at the earliest possible contact, improving quality of care and safety in acute care for individuals with dementia.

Introduction

In addition to creating risk to patients and hospital staff alike, behavioral and psychological symptoms related to dementia (BPSD) are associated with escalation of health-care costs and poorer health outcomes postacute care. ^{1 –3} Incidents of BPSD among individuals with Alzheimer’s disease or other dementias (ADOD), such as aggressive behavior, are challenging to respond to and may cause significant roadblocks in patient care in any treatment context. ^3,4 The objective of this research is to identify a screening indicator to health-care professionals who function in settings where historical record access may be at a minimum and rapid response to unknown patients may be required, such as in acute care and emergency department settings. Identifying individuals who are more likely to experience BPSD early in the intake process could be tremendously impactful for all involved. ⁵

Successful diversion of BPSD expressions begins with recognizing early signs, risk factors, and trigger events. ^3,5 Quickly identifying incoming patients who are at increased risk of BPSD will allow health-care professionals to alter modifiable variables, including their approach to the patient and some of the qualities of the environment. Prevention programs that focus on adapting the environment to lower stimuli that trigger challenging or dangerous behaviors can be effective for improving outcomes for medically vulnerable populations in general ⁶ and individuals with ADOD in particular. ⁷

While previous work identified some linkages to a documented history of mental health diagnosis other than dementia in known patients, ⁸ this research focuses on risk-stratification strategies gleaned from data in a retrospective record review in a large hospital system that might be implemented by medical personnel rapidly, in high-volume settings, and with unknown individuals presenting for care.

Background

White et al ⁴ measured that the overall prevalence of BPSD in an acute general medical hospital was 75% in patients living with ADOD, with aggression and activity disturbance being the most commonly reported concerns. The prevalence, however, tends to vary significantly between studies due to poor recognition of triggers, thresholds of severity, and setting. Brodaty et al ⁹ reported prevalence rates between 61% and 88% among the general population of people living with dementia in a community setting and 95% among hospitalized patients in acute care. Aggression, one of the most commonly reported and highest risk expressions of BPSD, is often preceded by pain, illness, sleep dysregulation, or overstimulation by environmental factors. ^{10 –12} Patients who come into an emergency department typically are greeted by an overwhelming barrage of new people, unfamiliar sights, considerable noise, intrusions into personal space, and unexplained procedures. Such factors are difficult for the patient with dementia to comprehend, are confusing and frightening, and may lead them to respond to this unknown threat through aggression, making aggression a common issue when combined with the medical condition that brought them there. ⁵ Due to the extreme vulnerability that patients living with ADOD exhibit in regard to their relationship with their environment, expressions of BPSD may result from the convergence of multiple, potentially modifiable, factors including internal (eg, pain, fear) and/or external (eg, overstimulating environment, complex caregiver communications) features. ^3,13

The prevalence of expressions of BPSD may be increased by interaction with staff who are not properly equipped or trained to care for a patient with dementia. Many hospital staff feel they do not possess the skills and confidence to work with “confused” patients. ¹⁴ The National Audit of Dementia Care in General Hospitals ¹⁵ found that only 5% of staff had received training in dementia care. Hospital staff often report that their clinics are understaffed, which can detract from the ability to properly manage patients with complex needs, such as behavioral disruption.

Relating Theoretical Frameworks With Clinical Approaches to BPSD in the Hospital Setting

Huesmann’s ¹⁶ information processing model has been applied to the analysis of BPSD among older adults with ADOD. Usually applied to youth, the model posits that interactions trigger assessment of situations and the social cues embedded in the context, and subsequent selection of scripts that govern affective and behavioral response. Theoretically, it can be inferred that individuals with impaired cognition would be unable to access the cognitive capacity to accurately assess and interpret social and affective cues in their environment. This would lead to a poor selection of scripts as a result of misinterpreted cues, leading to a greater likelihood of choosing inappropriate responses, such as aggression, to otherwise normal interactions. ⁸

Among other strategies focusing on medication management and staff training, ⁵ the progressively lowered stress threshold (PLST) model has emerged as a strategy for preventing BPSD among vulnerable patients. The PLST model cites fatigue, overwhelming or misleading stimuli (including noise level, social interaction, television, and/or mirrors), alterations to the patient’s normal context (environment, routine, or caregiver), excessive cognitive demand (including answering questions), perception of loss, and delirium as stressors that may precede episodes of increased behavioral expression. ^5,17

Although there is no accepted method to screen for challenging behavioral expressions or high-risk behaviors at intake, many strategies have been suggested. One strategy is to assume that patients who are more severely cognitively impaired or whose disease is further progressed are likely to exhibit more agitated behaviors in high stimulus environments, ¹⁸ although this approach is limited by the lack of linear progression of cognitive capacity and coping strategies in dementia. Additionally, such an approach overlooks the frequent presentation of delirium, which not only disrupts cognition but also may present with hallucinations or delusions that are distressing to the patient but tied only to the delirium and not to the larger presentation of dementia. ¹⁹ Although evidence suggests that hospital patients who have a history of premorbid personality problems are more likely to display aggression, this is not helpful as a screening method unless the patient is well known to the available staff, ²⁰ although it may be possible to glean such premorbid conditions in unknown patients through the presence of medications that are being currently prescribed. Higher levels of dependency upon caregivers and male gender have also been correlated with increased aggression, although the body of evidence regarding gender is equivocal. ^5,21,22 Furthermore, staff–patient interaction has been identified as the most likely potential triggering event for physical aggression by patients with ADOD; this is consistent with the PLST model regarding triggers. ^23,24 Given this information, we propose that there may be some individuals with ADOD who are more likely than others to become aggressive and that data may provide some insight into correlational relationships.

Because delirium is related to disrupted cognition and disorientation, we hypothesize that there is a relationship between individuals presenting to intake with a delirium and increased risk of aggressive behavior. Additionally, because of the established relationship between prescription of antipsychotic medications and BPSD in cases of dementia, ²⁵ we hypothesize that individuals who present to intake with dementia and who are taking a psychotropic medication will have increased risk of aggressive behavior while in the hospital and that an interactive effect of these variables may present a viable rapid screening option for increased risk.

Methods

Using the Electronic Medical Record Search Engine (EMERSE) search tool, ²⁶ we searched the electronic medical records of a large, Midwestern hospital. We identified adults aged 60+ who spent at least 24 hours at the hospital in any area except the psychiatric unit during the 1-year study time frame (N = 14 080). We sampled the first 4000 records and every 10th person through 14 080 who met these criteria (N = 5008; see Figure 1). The EMERSE is a web-based tool that was originally designed to search for patient-specific information from the University of Michigan Clinical Data Repository. Using EMERSE, investigators can search keywords without having to review all documents manually. Once records are identified as containing the search terms, the files are opened and reviewed manually. This study included 2 independent searches for incidents of dementia and incidents of physical aggression among the sample, then used the search engine to narrow the results within the searches by cross-searching (dementia within the aggression search and aggression within the dementia search), followed by a manual search of the narrowed field for incidents of delirium and medications at intake.

Figure 1.

Flowchart of data inclusion.

Dementia was identified in the electronic medical record by searching for an extensive string of search terms. Any notation that was made by or on behalf of a provider qualified to make such a diagnosis, or a notation of a suspected diagnosis, was counted as positive. Aggression was identified using a search string of behaviors such as violen*, fight*, slap, inappropr*, aggress*, combative, and so on and excluding terms such as “aggressively managed,” “aggressive treatment,” “punch biopsy,” and similar medical intervention-related terms. Delirium or notations of “altered mental status” were manually searched for in identified records and assessed by clinical research team members using criteria from the confusion assessment method, a validated screening tool for delirium, based on symptom descriptions in the case notes if no direct diagnosis was listed. Psychotropic medications were searched in identified records using EMERSE for an initial screen, then manually for detail. The EMERSE search used an extensive list of more than 200 terms that included common misspellings of some medications (eg, both paroxitine and paroxetene). Data entry was done by the primary researcher and a research assistant. Random error checks were conducted throughout the process, where each would randomly select entries of the other and check the data in the medical record for accuracy. Data were entered into and analyzed using SPSS (Version 23) and the data set was deidentified prior to analysis. The University of Michigan Health System institutional review board approved this study and data were reviewed by Compliance for protection of human patients (HUM00079867).

Analyses for this article examined only adults with ADOD (n = 505). Cases were dropped listwise by analysis. Data used for analyses are from either the admission with most recent incident of aggression or most recent admit within the time frame if no aggression occurred. We used logistic regression to examine relationships between aggressive inpatient behavior and gender, race, age at admit, delirium or “altered mental status” at intake, and whether the person was taking a psychotropic medication at intake.

Results

Average age for this sample (n = 505) was 78.2 years (standard deviation = 7.7 years). The sample was predominantly white (80.5%), 263 (52.3%) had a delirium diagnosis at admit, and 269 (53.5%) were taking a psychiatric medication at admit.

Analyses involved stepwise logistic regressions (see Table 1). In the first step, demographics, psychiatric medications at intake (yes/no), and delirium were entered. The second step included a term reflecting the interaction of delirium at intake and psychiatric medication prescription at intake. In the first step, the only significant correlate of aggressive behavior was psychiatric medication at intake (B = .82, standard error [SE] = .23, Wald = 13.22, P < .001, Exp(B) = 2.27). In the second step, it was found that the interaction between delirium and psychiatric medication at intake significantly related to incidence of aggression (B = −1.15, SE = 0.46, Wald = 6.15, P = .013, Exp(B) = 0.318), and there were significant main effects for both psychiatric medication at intake (B = 1.48, SE = 0.36, Wald = 16.56, P < .001, Exp(B) = 4.37) and delirium status (B = 0.93, SE = 0.38, Wald = 6.08, P = .014, Exp(B) = 2.53). Data suggest that there is a relationship between delirium at intake and aggression, or being on a psychotropic medication at intake and aggression, and that the presence of psychotropic medications at intake may moderate the effect of aggression risk associated with delirium.

Table 1.

Results of a Stepwise Logistic Regression Analysis Examining Aggression Risk.

	Step 1				Step 2
	B	SE	Wald	Exp(B)	B	SE	Wald	Exp(B)
Gender	−0.29	0.22	1.86	0.75	−0.28	0.22	1.62	0.76
Minority	0.31	0.27	1.37	1.36	0.30	0.27	1.29	1.35
Age	−0.004	0.01	0.07	1.00	−0.01	0.01	0.11	1.00
Delirium status	0.19	0.21	0.77	1.21	0.93	0.38	6.08a	2.53
Psychiatric medication	0.82	0.23	13.22b	2.27	1.48	0.36	16.57b	4.37
Delirium × Psychiatric medication					−1.15	0.46	6.15c	0.32
Constant	−1.36	1.12	1.49	0.26	−1.74	1.14	2.32	0.18

Abbreviation: SE, standard error.

^a P < .01.

^b P < .001.

^c P < .05.

A progression of χ² analyses was completed to elucidate the interaction. Those with a psychiatric medication prescription at intake were disproportionately female (χ² = 4.8, P = .03) and were at increased risk for aggressive behavior during their hospital stay (χ² = 12.4, P < .001). Patients with a psychiatric medication prescription at intake were not more likely to present with delirium (χ² = 0.4, P = .55). Among those who did not have delirium at admit, patients who were prescribed psychiatric medications were more likely to be aggressive than those who were not on psychiatric medication (χ² = 17.41, P < .001). Among those with delirium, there was not a significant difference between those who had been prescribed psychiatric medications and those who were not prescribed those medications with respect to aggressive behavior (χ² = .89, P = .35). Finally, the sample was divided by the presence of psychiatric medications (see Figures 2 and 3). Patients who were not prescribed psychiatric medications at admit were more likely to be aggressive if they had a diagnosis of delirium (χ² = 6.32, P = .01). Among those who had been prescribed psychiatric medication at admission, there was no difference in aggression rate between those who had been prescribed psychiatric medication and those who had not (χ² = .81, P = .37). In combination, findings indicate that psychiatric medications correlated with increased risk of aggressive behavior among dementia patients who did not have delirium, but had no effect on aggression risk among those who did have delirium.

Figure 2.

Delirium or psychiatric medication at intake and aggression: antipsychotic meds at intake (no).

Figure 3.

Delirium or psychiatric medication at intake and aggression: antipsychotic meds at intake (yes).

Discussion

Through examination of a 1-year cross section of medical records for older adults with dementia in an inpatient setting, we identified several consistent patterns. Individuals who behaved aggressively in the inpatient setting were more likely to have a delirium at intake. Additionally, and perhaps more easily identified in the fast-paced environment of an emergency or intake department, being on an antipsychotic medication at intake, regardless of reason, appears to be related to increased likelihood of aggressive behavioral expression during the inpatient stay.

Generalization of these findings should be made cautiously, as this was a single year sample from one hospital system. Additionally, although the sample size is sufficient for these analyses, we cannot be sure that this is representative of all patients that might have fit the inclusion criteria. The sampling method used was an attempt to mitigate any bias through systematic random sampling, but any method involves inherent limitations that include the unknown. We would point out that our analysis implies neither causation nor a universal screener. We do present a strong correlation resultant from a robust data set that may be used purely for clinical adaptation to lower risk and improve the care experience for particularly vulnerable patients in a normally busy and fast-paced setting. Although imperfect, it appears that recognition of possible delirium, coupled with an existing prescription for an antipsychotic medication at intake, may be a reliable means of screening for BPSD risk and intervening at the earliest possible contact. By identifying individuals at intake who are at a higher risk for aggressive behavior during their inpatient stay, it may be possible to strategically modify the environment and ways of interacting with the patient in such a way as to substantially decrease this likelihood.

Clinical Implications

Individuals living with dementia are particularly vulnerable in hospital settings. Disorientation, delirium, and stimulus factors that are both internal and external may trigger behavior that is difficult to manage or dangerous to staff. For individuals with dementia, there are many factors which may increase the risk of aggressive behaviors. Previous work identified a relationship between a known history of mental health diagnosis other than dementia and a propensity for disruptive expressions of behavior ⁸ ; this posed an interesting challenge, however, in applying the same principle to patients who may be unknown to a health-care system, or who may be arriving in fast-paced emergency departments. For those who are not able to access historical patient records, the ability to identify factors such as a medication prescription as a variable to consider in attempting to stratify risk propensity among incoming patients provides a unique opportunity to intercept those patients who would benefit from adaptations either in approach to their care or in the environment. The analyses presented here identify risk factors that are salient in such environments, providing opportunities for hospital personnel to gather information that allows them to be more rapidly prepared. Delirium and the presence of an antipsychotic medication at intake are highly correlated with an increased risk for aggressive behavioral expression while in the hospital. Identifying patients with dementia and either or both of these factors present at intake provides an opportunity to lower risk, increasing quality of care and patient safety for this population as well as for the hospital staff who care for them.