Psychotropic Medication Use and Changes During Hospitalization for Older Adults Living With Dementia

Abstract

To describe the use of psychotropic medications among older hospitalized patients. This was a descriptive study using baseline data from the first 308 older patients in a function focused care intervention study. Age, gender, race, comorbidities, admitting diagnosis, and medications (antidepressants, antianxiety medications, anticonvulsants, dementia drugs, antipsychotics, sedative-hypnotics, and opioids) were obtained at baseline and discharge. To compare change over time, generalized estimating equations were used. Participants were mostly female (63%) and White (69%) and were 83.1 years old on average. Antidepressant, antianxiety, anticonvulsant, dementia medication, sedative-hypnotic, and opioid use remained essentially unchanged between admission and discharge. Antipsychotic medication use increased significantly from 16% to 21% at discharge. There was persistent use of psychotropic medication among hospitalized older adults living with dementia and little evidence of deprescribing. There was some indication of changes made during hospitalization that may be appropriate, even without a focused deprescribing initiative.

Behavioral and psychological symptoms associated with dementia are common and include disturbances in perception, thought content, alterations in mood, or change in behavior. These symptoms are observed most commonly as agitation, aggression, anxiety, depression, repetitive behaviors, psychosis, resistiveness to care, apathy, wandering, alterations in sleep, and/or being verbally or physically inappropriate (Lyketsos et al., 2011). Psychotropic drugs are medications that alter mood, perceptions, and behavior and are used by many healthcare providers to treat the symptoms associated with dementia. Psychotropics include sedative/hypnotics, antidepressants, anticonvulsants, anxiolytics, antipsychotics, and opioids. These drugs are sometimes used off label and their use is not supported by evidence or guidelines (Maust et al., 2015, 2016). There are instances, however, in which these drugs are appropriate such as when an individual is experiencing distressing hallucinations or when he or she is a danger to self or others (Capiau et al., 2021; Kales et al., 2019). All the drugs in these groups can have a significant impact on cognitive function and subsequently on physical function and may result in sedation, confusion, delirium, falls, a decline in function, and death (Janus et al., 2017; Markota et al., 2016; Seppala et al., 2018). Due to the risk of adverse drug reactions and limited benefit, psychotropic medications are noted to be potentially inappropriate medications when prescribed for individuals living with dementia (Li et al., 2022).

Potentially inappropriate medications are those that have an unfavorable risk/benefit profile particularly when safer or equally effective alternative treatments are available (Prudent et al., 2008). It has been emphasized over the past decade to reduce the use of potentially inappropriate medications. In 1991, the Beers criteria was established to increase awareness of medications that were potentially inappropriate to use with older adults. This list has been revised through ongoing updates based on clinical findings (American Geriatrics Society, 2019). Another tool, the Screening Tool of Older Persons’ Potentially Inappropriate Prescriptions (Gallagher & O’Mahoney, 2008; Van der Linden et al., 2014) was also developed to help guide healthcare providers in stopping potentially inappropriate medications. Starting in 2005, the Federal Drug Administration issued a “black box” warning for use of atypical and typical antipsychotics (U.S. Food and Drug Administration, 2005). To further focus on reducing antipsychotic use in nursing homes, the Centers for Medicare and Medicaid Services initiated the National Partnership to Improve Dementia Care in Nursing Homes in 2012 (Centers for Medicare and Medicaid, 2012). Other initiatives have encouraged this focus on decreasing psychotropic medications such as the Choosing Wisely campaign initiated by the American Board of Internal Medicine (American Board of Internal Medicine, 2012).

Despite these guidelines and recommendations, the use of potentially inappropriate medications, specifically psychotropic medications, is common in hospitalized older adults. Earlier studies published in 2006 to 2008 noted that approximately 40% to 50% of hospitalized older adults were on at least one psychotropic medication (Barry et al., 2006; Gallagher et al., 2008; Prudent et al., 2008). The rate of use continued to be high in 2018 with 46% to 58% of older patients admitted to acute care receiving a psychotropic medication and this increased to 72% by discharge (Adeola et al., 2018; Silwanowicz et al., 2017). Most recently published research has shown rates of psychotropic use among hospitalized older adults to be at least 40% during the hospital stay (Lam et al., 2019; Li et al., 2022).

Conversely there has been some research focused on deprescribing of potentially inappropriate medications among older adults in acute care (Drago et al., 2020; Thillainadesan et al., 2018). The programs tested included use of electronic health records as prompts and alerts, pharmacist-led interventions, prescriber education programs, multidisciplinary interventions, and clinical decision support systems (Thillainadesan et al., 2018). In a systematic review (Thillainadesan et al., 2018) of these interventions, it was concluded that the strategies used were feasible and effective at reducing potentially inappropriate medication use and did not cause harm. The hospitalization, therefore, should be considered an opportunity for initiating and monitoring a decrease in dosage or discontinuation of potentially inappropriate medications. The advantage of initiating this in acute care settings is having 24-hr monitoring by registered nurses. This registered nursing oversight is not available in most nursing homes, assisted living communities, or home settings.

Study Purpose

The purpose of this secondary data analysis was to build on prior research and describe the current use of psychotropic medications during the hospital stay among older patients living with dementia. Data from the study, Testing the Impact of the Function Focused Care for Acute Care Using the Evidence Integration Triangle Intervention (FFC-AC-EIT), were used. It was hypothesized that controlling for age, gender, race, comorbidities, cognition, and hospital site (exposure or not to the Function Focused Care Intervention), there would be a decrease in use of psychotropic medications from admission to discharge. The findings from this study can help inform researchers and clinicians about the current management of psychotropic medication use among these patients.

Methods

Design

This was a secondary data analysis using baseline data from the first 308 participants in a cluster randomized controlled trial testing the effectiveness of FFC-AC-EIT (anonymized for peer review). Participants were from the first ten study hospitals located in two states with five randomized to FFC-AC-EIT and five randomized to function focused care education only. Supplemental Appendix 1 provides an overview of the intervention. The study was approved by a university-based Institutional Review Board.

Settings

Hospitals were invited to participate in the study if they had the following: (1) at least one unit dedicated to general medical patients; (2) two nurses who were willing to be champions (one for day shift and one for evening shift); (3) email and website access for staff via a phone, tablet, or computer; and (4) not already established a unit specifically focused on optimizing care provided to older adults.

Sample

Patients were eligible to participate if they were (1) 55 years of age or older; (2) admitted with a medical versus surgical diagnosis (excluding patients with COVID-19); and (3) cognitively impaired based on results from all of the following cognitive measures: a score ≤20 on the Saint Louis University Mental Status (SLUMS) exam for those with a high school education or ≤19 for patients with less than high school education (Morley & Tumosa, 2002); >2 on the AD8 Dementia Screening Interview (Galvin et al., 2005); 0.5 to 2.0 on the Clinical Dementia Rating Scale (CDR) (O’Bryant et al., 2008); and a score of ≥9 on the Functional Activities Questionnaire (Pfeffer et al., 1982). All cognitive measures were completed based on input from legally authorized representatives, with the exception of the SLUMS, which was completed by the patient. Patients were ineligible if they (1) were receiving Hospice services; (2) were on the unit for greater than 48 hr; (3) had no individual identified to contact for follow up; (4) were being evaluated for surgery; or (5) had evidence of a major acute psychiatric disorder or significant neurological condition associated with cognitive impairment other than dementia. All approached patients had to pass an evaluation to sign consent (anonymized for peer review). If they did not pass this evaluation, then the legally authorized representative was contacted to complete the consent process.

To obtain the sample, 4,643 patients were screened and 2,854 patients were eligible to be approached with the majority of those ineligible for approach being due to testing positive for, or suspected of, having COVID-19. We consented 556 individuals (14% of those who were eligible) and enrolled and randomized 313 (11%) individuals as the other consented individuals were not eligible mainly due to cognitive eligibility (they did not screen positive for dementia). The sample was mostly White women although there were 37% males and 30% Black or non-White participants. A total of 308 participants had admission and discharge medication data and were included in the study.

Procedure

Study research evaluators obtained descriptive data and medications from the electronic medical records at baseline and discharge. The evaluators all had prior experience collecting this type of data.

Measures

Descriptive data included age, gender, race, comorbidities, admitting diagnosis, and medications. The SLUMS (Morley & Tumosa, 2002) was used as the measure of cognitive status. A score of ≤20 on the SLUMS exam for those with a high school education or a score of ≤19 for patients with less than high school education provides evidence of cognitive impairment (Morley & Tumosa, 2002). Psychotropic medications were obtained and included the following classes of drugs: antidepressants, antianxiety medications, anticonvulsants, dementia drugs (i.e., cholinesterase inhibitors and memantine), antipsychotics, sedative-hypnotics, and opioids. Dose and frequency were also recorded. Drugs ordered at baseline (within 24 hr of recruitment) and discharge (within 24 hr of discharge) were obtained from the electronic health record.

Data Analysis

Descriptive statistics were done using the Statistical Package for the Social Sciences version 27.0 to describe the sample and medication administration using SPSS 27.0. To compare changes in medication, use over time generalized estimating equations were used. For each outcome, exploratory analyses (tests for normality, scatterplots, frequencies, and boxplots) were performed to assess model assumptions. In all analyses age, gender, race, comorbidities, cognition, and treatment group were controlled for. The Wald statistic was used, and all tests were two-sided with a 5% significance level and were adjusted for clustering within settings.

Results

Table 1 provides descriptive findings of the sample. The participants were mostly female (63%) and White (69%) with an average age of 83.1 years. There were only 3 (1%) Hispanic participants and most participants had at least a high school education (84%). An average of 2.0 (SD = 1.3) comorbidities were reported and the overall cognitive status based on the SLUMs was 7.1 (SD = 6.1), which indicates that there is evidence of dementia. The reasons for admission varied with the greatest percentage being for infections (28%) and then for cognitive, behavioral, or functional changes (13%).

Table 1.

Sample Descriptives.

Variable	N (%)	Mean (SD)
Gender
Male	113 (36)
Race
White	215 (69)
Ethnicity
Hispanic	3 (1)
Education
<High school	49 (16)
High school	150 (49)
Some college	38 (12)
Associate/bachelor’s degree	49 (17)
Graduate school	31 (6)
Marital status
Single	39 (13)
Married	111 (36)
Separated	1 (1)
Divorced	27 (9)
Widowed	130 (41)
Age		83.1 (7.9)
Comorbidities		2.0 (1.4)
Cognition (SLUMS score)		7.8 (6.1)
Reason for admission
Infection	87 (28)
Change in cognition	41 (13)
Musculoskeletal disease	33 (10)
Cardiac disease	29 (9)
Neurological disease	24 (8)
Other	18 (6)
Respiratory issues	17 (6)
Gastrointestinal issues	14 (5)
Kidney damage	15 (5)
Electrolyte imbalance	12 (4)
Anemia/bleeding	11 (4)
Diabetes	6 (2)

Table 2 provides a description of the number and percentage of participants prescribed any of the medications within the group (e.g., any antidepressant) on admission and discharge and the number and percentage of participants in which a drug within the group was started or stopped during hospitalization. Table 3 provides the specific drug used and started or stopped during the hospital stay. As shown in Table 4, aside from antipsychotic medication use, there was very little significant change in the use of medications within each drug group between admission and discharge. Overall, 38% of the participants were on antidepressants on admission and 37% at discharge (Table 4, Wald = .36, p = .55). Thirteen percent of the participants were on antianxiety medications on admission and 12% at discharge (Wald = 2.01, p = .16). Regarding anticonvulsants, 23% were on at least one drug within this group on admission and this increased to 24% at discharge (Wald = .40, p = .53). Similarly, dementia medication use increased nonsignificantly from 17% on admission to 18% on discharge (Wald = 1.61, p = .21). Antipsychotic medication use increased significantly from 16% to 21% at discharge (Wald = 10.18, p = .001). Sedative-hypnotics remained the same at 4%, and opioid use showed a nonsignificant decrease from 19% to 18% at discharge (Wald = 2.86, p = .09).

Table 2.

Drug Use Admission and Discharge and Change Between Admission and Discharge.

Drug group	Medication use on admission N (%)	Medication use at discharge N (%)	Medication stopped on admission N (%)	Medication started after admission N (%)	Medication unchanged admission to discharge N (%)
Antidepressants	117 (38)	114 (37)	14 (5)	11 (4)	103 (33)
Antianxiety	42 (14)	36 (12)	12 (4)	6 (2)	30 (10)
Anticonvulsants	72 (23)	74 (24)	4 (1)	6 (2)	68 (22)
Dementia	53 (17)	57 (18)	3 (1)	7 (2)	50 (16)
Antipsychotic	50 (16)	66 (21)	5 (1)	21 (7)	45 (15)
Sedative–hypnotic	13 (4)	13 (4)	1 (.5)	1 (.5)	12 (4)
Opioid	60 (19)	55 (18)	17 (6)	12 (4)	43 (14)

Table 3.

Specific Medication Used and Changes During Hospital Stay.

Medication used	Total treated	Discontinued during admission	Started during admission
Antidepressants
Venlafaxine	2	0	2
Escitalopram	4	1	3
Mirtazapine	8	6	2
Duloxetine	4	2	2
Sertraline	5	2	3
Fluoxetine	1	1	0
Paroxetine	1	0	1
Triazolopyridine	2	1	1
Total	27	13	14
Antianxiety
Buspirone	1	1	0
Lorazepam	12	9	3
Hydroxyzine	2	0	2
Alprazolam	2	2	0
Diazepam	1	0	1
Total	18	12	6
Anticonvulsants
Gabapentin	7	5	2
Divalproex sodium	1	0	1
Levetiracetam	1	0	1
Topiramate	1	0	1
Zonisamide	1	1	0
Total	11	6	5
Dementia
Memantine	5	2	3
Donepezil	5	3	2
Rivastigmine	1	0	1
Total	11	5	6
Antipsychotics
Prochlorperazine	1	0	1
Haloperidol	1	1	0
Droperidol	7	2	5
Olanzapine	1	0	1
Quetiapine	10	1	9
Risperidone	6	1	5
Aripiprazole	1	0	1
Total	27	5	22
Opioids
Oxycodone	6	2	4
Tramadol	11	7	4
Morphine	8	6	2
Hydrocodone	2	1	1
Hydromorphone	1	1	0
Butalbital/acetaminophen/caffeine	1	0	1
Total	29	17	12

Table 4.

Change in Medication Use Between Admission and Discharge.

			Unadjusted total group over time		Compared by treatment group and adjusted*
Medication group	Admission N (%)	Discharge N (%)	Wald Statistic	F(p)	Wald Statistic	F(p)
Antidepressant			.001	.39 (.53)	.001	.003 (.96)
Yes	117 (38)	114 (37)
No	192 (62)	195 (63)
Antianxiety			.008	.22 (.64)	.001	.28 (.60)
Yes	42 (13)	36 (12)
No	267 (87)	273 (88)
Anticonvulsant			.001	.22 (.64)	.013	2.86 (.09)
Yes	72 (23)	74 (24)
No	237 (77)	235 (76)
Dementia			.003	.77 (.38)	.007	1.65 (.20)
Yes	53 (17)	57 (18)
No	256 (83)	252 (82)
Antipsychotic			.029	8.77 (.003)	.012	2.74 (.10)
Yes	50 (16)	66 (21)
No	259 (84)	243 (79)
Opioid			.004	1.32 (.25)	.007	1.51 (.22)
Yes	60 (19)	55 (18)
No	249 (81)	254 (82)
Sedative-hypnotic			.00	.00 (1.00)	.001	.033 (.86)
Yes	13 (4)	13 (4)
No	296 (96)	296 (96)

^*Adjusted for age, gender, cognition, race, and comorbidities.

Regarding antidepressants, as shown in Table 3, mirtazapine was the only drug stopped in a greater number of participants than it was started (stopped on six patients and started on two). For antianxiety medications buspirone, lorazepam, and alprazolam were all stopped in a greater number of participants than they were started. For anticonvulsant medications, gabapentin and zonisamide were stopped in a greater number of participants than started. For dementia medication, donepezil was the only medication that was stopped in a greater number of participants than it was started. For antipsychotics, haloperidol was the only medication stopped in a greater number of participants than it was started. Conversely, quetiapine was started in nine patients and only stopped in one, and risperidone was started in five patients and only stopped in one. Lastly, regarding opioids, a greater number of individuals had tramadol, morphine, and hydromorphone stopped versus an increase in the number of patients prescribed oxycodone and butalbital/acetaminophen/caffeine.

There was very little evidence of deprescribing in terms of decreasing dosages of medications. There were two cases in which gabapentin dosage was decreased, one case in which tramadol dosage was decreased, and one case in which phenytoin dosage was decreased.

Discussion

The stated hypothesis that there would be a decrease in the use of psychotropic medications from admission to discharge was not supported. Instead, the findings from this study noted there was continued use of psychotropic medications among the participants. Moreover, there was little evidence of attempts to decrease use of or decreased dosages of psychotropics during the hospital stay. Rates of use of antidepressants and antianxiety medications were consistent with or slightly lower than that reported in prior studies (Adeola et al., 2018; Lam et al., 2019; Prudent et al., 2008). The rate of antipsychotic use varied greatly across prior studies with ranges from 4% in a study done in Australia (Lam et al., 2019) to as high as 35% in a study at Taipei Medical University Hospital (Li et al., 2022). The current study fell within that range at 16% on admission to 21% at discharge. Sedative-hypnotic use in the current study was 4%, which was generally lower than that noted by prior research report ranges from 7% to 35% (Adeola et al., 2018; Lam et al., 2019; McDonald et al., 2019; Prudent et al., 2008; Zelinkova et al., 2021). The prior papers did not focus specifically on older adults living with dementia and thus dementia related medications were not addressed. These drugs are important to consider due to their risk benefit profile and potential cholinergic side effects such as gastrointestinal disturbances and fatigue. Differences in rates of medication are likely due to differences among samples, the time frame of the studies, dissemination of new drug information, new medication options (e.g., increase in second generation antipsychotics), and philosophies of care across the different geographic locations and healthcare providers.

This study was a secondary data analysis and the reasons for medication changes were not collected. Therefore, it is impossible to know if the changes made during the hospital stay were appropriate. This is particularly relevant for the increased use of antipsychotic medications. These drugs may have been used appropriately, inappropriately, and/or off label to manage behavioral symptoms with little evidence to support their effectiveness (National Institute for Health and Care Excellence, 2015; Rosenberg et al., 2012). It is possible that drugs were discontinued due to side effects such as weakness, sleepiness, or delirium/change in cognition. Further drug changes may have been based on current recommendations for use of certain antipsychotics over others (e.g., use of quetiapine vs. haloperidol due to side effect profile). Future research should focus on obtaining documentation for initiation, changes, and outcomes associated with psychotropic medication for hospitalized patients living with dementia. This information could then be shared in discharge summaries so that subsequent care settings and healthcare providers can focus on appropriate medication use (Adeola et al., 2018; Lam et al., 2019; Li et al., 2022).

This descriptive secondary data analysis used data from the study testing the impact of FFC-AC-EIT which was focused on increasing physical activity among patients living with dementia when hospitalized. The study did not include an intervention focused on deprescribing psychotropics. Prior research, including a systematic review, has shown that deprescribing interventions were feasible and did not cause harm (Drago et al., 2020; Thillainadesan et al., 2018). The studies included, however, were of low quality and the impact of deprescribing on clinical outcomes (e.g., quality of life) is still not clear. Research needs to focus on this particularly for those living with dementia as the pros and cons of psychotropic medication are very individualized (Alliance for Aging Research, 2022).

Conclusion

The descriptive findings from this study suggest that there was persistent use of psychotropic medication among hospitalized older adults living with dementia and little evidence of deprescribing. Based on guidelines such as the Beer’s Criteria and known drug side effects, there were some positive changes made in medication use for study participants. Specifically, mirtazapine, buspirone, lorazepam, alprazolam, gabapentin, zonisamide, and donepezil were stopped in more patients than started during the hospitalization. The hospital is an ideal location to initiate medication changes. Future research and clinical care should focus on exploring the diagnosis and rationale for prescribing and deprescribing decisions and the impact or outcome of these changes at discharge and over time.

Study Limitations

This study was limited in that it was a secondary data analysis and did not include information about the rationale for drug use or changes. Data were not collected for “as needed” medications, frequency of administration, or patient response to the medication. Furthermore, this was a small local sample from just two states. The study was done during the COVID-19 pandemic, and medication management may not have been a priority for healthcare providers during this period. Despite these limitations, the data provide current descriptive findings about psychotropic medication use among hospitalized older adults living with dementia and can be used to help guide future clinical work and research.

Biographies

Author Biographies

Barbara Resnick, PhD, CRNP is a Professor and Sonya Gershowitz Chair in Gerontology at the University of Maryland School of Nursing.

Marie Boltz, PhD, CRNP, is a professor and the Elouise Ross Eberly and Robert Eberly Endowed Chair at the Penn State School of Nursing.

Elizabeth Galik, PD, CRNP, is a professor and Chair of the Department of Organizational Systems and Adult at the University of Maryland School of Nursing.

Ashley Kuzmik, DrPH, MPH, is a post doctoral student at the Penn State University, School of Nursing.

Brittany Drazich, PhD, MSN, is a post doctoral student at the University of Maryland School of Nursing.

Rachel McPherson, PhD, is a post doctoral student at the University of Maryland School of Nursing.

Nayeon Kim, BSN, is a pre doctoral student at the University of Maryland School of Nursing.

Chris Wells, PhD, PT, CCS, FCCM, is a physical therapy at the University of Maryland Medical System.

Shijun Zhu, PhD, is a Professor and statistician at the University of Maryland School of Nursing.