Improving Care through Improv: Pilot Test Results of an Innovative Training Program for Care Partners of People Living with Dementia

Abstract

Background:

Identifying effective evidence-based ways to prepare the growing number of family members and friends who are or will be providing care to someone living with dementia is a critical public health issue. Training care partners in skills associated with improvisational (improv) theatre holds promise for improved communication yet requires formal investigation.

Design and Methods:

We pilot tested the feasibility, acceptability, and preliminary efficacy of the training program, “Improving Care through Improv,” among care partners of people living with moderate dementia. We used a non-controlled trial design with participants surveyed at three time points. Care partners received in-person training co-designed and co-led by an improv educator and a gerontologist during weekly 2-hour group sessions held over 4 weeks. Surveys assessed caregiver mastery (primary outcome) and self-perceived stress, anxiety, caregiver burden, perceived ability to manage behaviors, and care recipient’s quality of life at baseline; 1-month (program completion), and 3 months post baseline.

Results:

Forty-three care partners participated, 42 completed the training, and 40 completed all intervention components. Dependent sample t-tests between times 1 and 3 showed improvement in caregiver mastery as well as reductions in perceived stress and depression. Three months post baseline, 100% reported using the skills taught in the training.

Implications:

Recruitment, enrollment, and completion rates indicate high levels of feasibility and acceptability. Findings demonstrate improv’s promise for improving dementia care experiences, illuminate areas for refinement and application, and support further testing.

Background and Objectives

Nationwide, over 12 million Americans, mostly family members and friends, provide informal unpaid care to millions of people living with dementia, care that frequently takes place over many years and is associated with unique challenges that escalate in number and intensity (Alzheimer’sAssociation, 2025). The nature of neurodegenerative disorders such as dementia, including Alzheimer’s disease, means that as the condition progresses beyond early stages, care partners typically provide greater and more diverse types of support. Along with requiring more assistance with activities of daily living and supervision (Wang et al., 2019), people living with moderate-stage dementia may experience increasing confusion and, depending on dementia type, memory loss, as well as significant changes in personality and behaviors. Some may also experience, for instance, distrust, paranoia, visual or auditory hallucinations, and have difficulty communicating and with expression, restlessness, anxiety, agitation, depression, and aggression (Alzheimer’s Association, 2025; James et al., 2020). Dementia care, especially for those living in more moderate stages, can involve navigating gaps in memory that are surprising and unexpected or changes in behaviors that are out-of-character (see Taylor, 2008). These changes often are among the most difficult for care partners to navigate (Feast et al., 2016; Ornstein & Gaugler, 2012; Vaingankar et al., 2016).

Most informal care partners assume their care roles with little, if any, training and are ill- or under-prepared for the challenges, especially the unpredictability typically associated with caring for someone living with dementia and entering the moderate stage of dementia (Brodaty & Donkin, 2009). Not surprisingly, as care roles intensify, care partners can experience accompanying threats to their well-being including increases in stress, burden, anxiety, and depression; physical health and mortality also can be compromised (Vaingankar et al., 2016). Given the interdependent nature of care relationships (Kemp, 2021), negative care partner outcomes also can have an equally deleterious effect on care recipients, compromising their quality of care and life, including the ability to avoid or delay nursing home placement (Villars et al., 2022). Bolstering informal care partners’ ability to succeed in their care roles and minimize negative outcomes is a critical public health issue with significant social and economic implications for individuals, communities, and nations. Needed are evidence-based interventions that equip family members and friends for the care role and improve their lives and the lives of people living with dementia.

This article reports on a pilot study designed to refine and test an in-person group-based psychoeducational program, “Improving Care through Improv.” This novel training incorporates fundamentals of improvisational (improv) theatre skills with practical knowledge about dementia and being a care partner. Supported by Emory’s Roybal Center for Caregiver Mastery, which is funded by the National Institute on Aging, this study assessed the feasibility, acceptability, and efficacy of the training to enhance caregiving mastery and well-being outcomes among care partners of community-dwelling people living with moderate dementia.

Conceptual Backdrop

Our intervention is informed by Social Cognitive Theory (Bandura, 1977, 1989, 2001), which suggests that situation-specific stresses and challenges can be overcome if people acquire appropriate knowledge and skills and develop relevant coping strategies. Bandura (1977, p. 193) underscores the role self-efficacy plays in coping, noting, “expectations of personal mastery affect both initiation and persistence of coping behavior. The strength of people’s convictions in their own effectiveness is likely to affect whether they will even try to cope with given situations.” Individuals’ beliefs in their capabilities influence how they anticipate and ultimately experience and manage stressful situations (Bandura, 1977), including caregiving.

Caregiver mastery, comprised of feeling competent and confident in the care role, is key to successful caregiving and, as research shows, can be learned. Perhaps as a result, psychoeducational programs are the most common type of dementia caregiver intervention (Encinas-Monge et al., 2024; Gilhooly et al., 2016). These intervention teach cognitive and behavioral skills with emphasis on understanding dementia, the complexities of dementia care, and the care role and identifying strategies to manage stressful situations (Frias et al., 2020). Effective evidence-based psychoeducational programing has been shown to reduce self-reported stress among dementia caregivers (Gallagher-Thompson et al., 2012). Randomized controlled trials (RCTs) consistently show decreases in caregiver burden and depression among those in intervention groups relative to groups receiving usual care (Encinas-Monge et al., 2024). The Savvy Caregiver Program (Savvy) and later Tele-Savvy, designed by Hepburn and colleagues to promote caregivers’ ability to apply solution-focused coping behaviors to daily care-related stressors, for instance, showed statistically and clinically significant benefits in enhancing caregiver mastery and improving caregiver well-being (Hepburn et al., 2007; Hepburn et al., 2022; Hepburn et al., 2001). Relative to pharmacological strategies, psychoeducational programs such as these tend to be safer and more effective in preparing caregivers to: navigate the complexities of care, better manage stress and coping, including responses to behavioral changes, and improve caregiver and recipient quality of life and well-being (Sun et al., 2022).

The Promise of Improv

Myriad disciplines from business to law, medicine, nursing, and other health professions are turning to improv training to improve flexibility, communication, collaboration, and creative thinking (Chan et al., 2023; Gao et al., 2019; Harding, 2020; Higgins & Nesbitt, 2021; Krueger et al., 2019; Robbins Dudeck & McClure, 2018, 2021; Zelenski et al., 2020). Improv typically involves performances that are largely unscripted, unplanned, and collaborative. An “applied improvisation mindset” encourages co-creation, accepting and building on what others are saying or doing without a need to correct or contradict (i.e., “Yes, and…), being in the moment, self-awareness, attentiveness and responsiveness to others’ verbal and non-verbal actions, giving up control, making your partner look good, leaning into uncertainty, accepting failure, and pivoting (Robbins Dudeck & McClure, 2018, 2021; see also Taylor, 2008).

Research shows that improv training can be used to address anxiety, enhance divergent thinking skills, and improve tolerance of uncertainty in social encounters (Felsman et al., 2020). Among health care practitioners and those in training, research suggests that improv education can enhance empathy, flexibility, observation and listening skills, and collaboration (Chan et al., 2023; Gao et al., 2019; Higgins & Nesbitt, 2021; Zelenski et al., 2020). Mehta et al. (2021) posit that improv training builds competence and confidence (i.e., mastery) among medical doctors.

Learning to successfully navigate dynamic and difficult care situations and embracing such uncertainty are critical to achieving mastery in the dementia care context and require improvisation (Kemp et al., 2025). Being prepared for the unexpected and having the ability to pivot when strategies, including previously successful ones, fail to work, are essential in dementia care because many care partners find that they “just never know what” they’re “gonna get from day-to-day” (Kemp et al., 2024, p. 304). Research bears out that flexibility, adaptability, and creativity – key improv tenets – are important and teachable skills in dementia care and can enhance care experiences and outcomes (Hepburn et al., 2003; Hepburn et al., 2001; Kemp et al., 2025; Kemp et al., 2024). Yet, unlike traditional educational programing that may look and feel like training or involve passive methods or self-directed learning, improv training involves social engagement, play, active embodied learning, all of which promote personalization, retention, and skill development (Bayne et al., 2021) giving improv-based programing potentially broader appeal to a greater number of care partners.

Improv and Dementia Care

Our pilot study was informed, in part, by two previous studies that designed and delivered improv training to dementia family caregivers: “Improv for Care” (Brunet et al., 2021) and “Improv to Improve” (Howell et al., 2022). Both interventions involved weekly 2-hour in-person group sessions delivered over six weeks and collected data on caregiver’s emotional well-being and care recipients’ neuropsychiatric symptoms before and after the training. Building on Krueger et al.’s (2019) work with therapeutic improv groups to treat anxiety and depression among people with mood disorders, Brunet (2021) led the first intervention. The team enrolled 29 family caregivers; 15 completed surveys pre- and post-intervention. Upon training completion, caregivers reported decreases in depressive symptoms, burden, and stress and increases in the frequency of their person’s neuropsychiatric symptoms, but not in associated distress. The team speculates that the intervention may have mitigated the relationship between symptoms and stress “by providing coping skills and social support” (p. 370).

Conducted by Howell and colleagues (2022), the second intervention was informed by research demonstrating the importance of respectful, flexible, and creative communication within dementia care (e.g., Basting, 2013) noting, as our own research confirms, that family members do not always start their caregiver journey with these communication skills or ever acquire them (Kemp et al., 2021; Kemp et al., 2023; Kemp et al., 2024). This team used Pearlin et al.’s (1990) stress and coping model as a guiding conceptual framework. With a focus on “improving listening and communication mastery,” their intervention sought to “promote positive coping and communication skills” among family caregivers of loved ones with dementia (Howell et al., 2022). Fifteen family members participated; 5 completed surveys. Caregivers experienced elevated mood and decreases in reports of their loved one’s neuropsychiatric symptoms and caregiver’s associated distress. Except for the decrease in reported symptoms, findings are consistent with those of Brunet and colleagues. However, the small sample size limits interpretation.

Ultimately, both studies point to the potential benefits of improv training on care partners’ well-being and for strengthening care interactions. Both identify the need for skill development and coping among caregivers and identify mastery as a potentially important explanatory concept. Additional development and testing among caregivers of people living with dementia, especially with larger sufficiently powered sample sizes, are warranted.

Successful Engagement Strategies and the Need for Improv-Based Training and Evaluation

Our recently concluded longitudinal study focused on assisted living residents with dementia (R01AG062310) also informs the improv programing at the center of this manuscript. Described extensively elsewhere, this ethnographic study sought to identify best practices for promoting meaningful engagement and quality of life for people living with dementia (Kemp et al., 2025). Analysis of fieldnote and interview data identified four approaches used by care partners, including staff, residents’ family members and friends, volunteers, external workers, and others, who exhibited exceptional capacity to successfully engage residents with dementia (Kemp et al., 2021; Kemp et al., 2023). These employed, effective approaches included: knowing the person, being in the moment, meeting people where they are, and viewing each encounter as opportunity (Kemp et al., 2024); all overlap with improv techniques.

Additional analysis from this project examined care partner responses to dementia-related confusion about time, place, people, and scenario (Kemp et al., 2025). Care partners engaged in “reality management,” responding in variable ways that invalidated or validated residents’ realties and emotions. First, ignoring or dismissing a resident’s reality led to frustration, even anger, and social withdrawal. Next, correcting could be harmful (e.g., being reminded of a forgotten painful loss) or beneficial (e.g., being reminded of something positive). Meanwhile, redirecting often involved distraction and was most successful when used with validation (e.g., “I would be upset too if I couldn’t find my purse. Why don’t we have lunch first and then I’ll help you look for it?”). Finally, some caregivers responded to residents’ confusion by joining or stepping in to their realities including, for instance, when residents believed that they were at work, in a hotel, or needing to go to school. Joining their reality closely aligns with improv. Although not always possible, the listening, patience, collaboration, and empathy that joining requires are essential in creating positive care interactions. Improv training, as we demonstrate, teaches these skills and has potential for building dementia care skills and competencies and improving care interactions. Most care partners did not naturally possess or use these techniques, reinforcing the need for further development and testing of improv-based care partner training (Kemp et al., 2024), including our program, “Improving Care through Improv.”

Improving Care through Improv

Our training program, informed by the research reported above, originates from a collaboration between improviser and educator, Amanda Lee Williams and the project’s principal investigator, Candace L. Kemp and partnerships with Dad’s Garage Theatre Company, Emory University, and Georgia State University (GSU). In 2022, Williams and Kemp developed the first version of the training as a 1-hour presentation and a 2-hour workshop focused on the delivery of dementia-friendly care for formal, paid staff of the geriatric workforce (Georgia Gear, 2019–2024). In 2023, Williams and Kemp expanded and adapted the “Improving Care through Improv” curriculum for family and friends. We presented the program to an expert advisory team comprised of dementia care partners/advocates who gave feedback iteratively through two workshops and one listening/feedback session.

“Improving Care through Improv” for family and friends is designed to augment care partner mastery. The training is an 8-hour group training program that builds care partner skills cumulatively over a series of four 2-hour in-person class sessions co-taught by an improviser educator and a gerontologist with expertise in dementia and dementia care. Training seeks to improve care partner competence and confidence in their role through the acquisition and application of knowledge and skills. Specifically, the training provides practical information about dementia and being a care partner for someone living with dementia, including for example, understanding different dementia types, tips for managing agitation, and the importance of self-care. The training also introduces foundational improv techniques and ways of thinking, communicating, and behaving. Classes facilitate skill development and the application of an improvisation mindset, including for instance, accepting invitations to engage, giving up control, and expecting the unexpected, in the dementia care context. The program involves instruction, demonstration, practice, application, reflection, and discussion. Table 1 outlines the organization of the program and curriculum by week. Following convention Ball et al. (2002) and for practical reasons, including the often unpredictability of dementia care partners’ lives, completing 3 of the 4 training sessions (or 75% of the material) constituted program completion.

Table 1.

Improving Care through Improv: Weekly Activities, Foci, and Learning Strategies

Week	Improv Foci	Dementia Care Foci	Learning Strategies
1	Introduction Course overview and purpose Team- and trust-building Introduction to improv Actively listening and observing Offering and accepting invitations	Understanding different types of dementia Tips for managing agitation, aggression, and sundowning Local dementia care resource Care connections	Learning circle – introduction Improv warm-up exercises Debrief Reviewing resources Improv team-building exercises Debrief Learning circle – reflection and sharing
2	Participant sharing Listening Staying present/being in the moment Communication strategies	Emotion management The importance of self-care Self-care strategies Care connections	Learning circle – check in/ application reflection Improv warm-up exercises Debrief Improv mindset exercises Debrief Emotion management and self-care discussion Improv communication exercises Debrief Learning circle – reflection and sharing
3	Participant sharing Storytelling Collaborating Making your partner look good Introduction to direct application	The importance of meaningful engagement Successful approaches to promote meaningful engagement Care connections	Learning circle – check in/ application reflection Improv warm-up exercises Debrief Improv collaboration exercises Debrief Dementia care strategies and application discussion Learning circle – reflection and sharing
4	Participant sharing Giving up control Pivoting and being flexible Expecting the unexpected Application to real care scenarios Final reflection	Person-centered care competencies Navigating different viewpoints and realities Applying the improv mindset Care connections	Learning circle – check in/ application reflection Improv warm-up exercises Debrief Improv application Debrief Improv care partner role play Debrief Learning circle improv application and take-aways

Each week, care partners receive a packet of materials comprised of core content, helpful resources, and homework tasks, including practice and optional journaling topics. Classes are two hours in length.

Method

Research Design

This one-year study employed a non-controlled trial design. Participants received the same 4-week training program and completed surveys at three time points: (T1) baseline/pre-training; (T2) Month 1/immediately upon training completion; and (T3) Month 3. We also used qualitative methods to better understand the study’s feasibility, acceptability, and efficacy. GSU’s Institutional Review Board approved the study (protocol H23627).

Recruitment

We used multiple recruitment strategies and convenience sampling to identify volunteers to pilot test “Improving Care through Improv.” First, Emory’s Integrated Memory Care program shared study information with patients’ family members via email, newsletter updates, and flyers. Next, we promoted the study through our expert advisory team members, professional networks, and social media channels, and with flyers posted in local senior centers, at Dad’s Garage Theatre, and on GSU’s campus. Finally, once enrollment began, we used snowball sampling; several participants referred the study to care partners in their networks.

Potential participants contacted the research team via email or phone to obtain study details and information on eligibility screening. Inclusion criteria included being: 1) 18 years of age or older; 2) able to understand and communicate in English; and 3) an informal care partner to a community-dwelling person with moderate dementia (a score of 4 “decreased ability to manage complex tasks” or 5 “assistance needed with clothing and some daily tasks” on the 7-point Functional Assessment Staging (Sclan & Reisberg, 1992). Potential participants were excluded if they did not meet these criteria or anticipated moving their person to a skilled nursing facility within the next 6 months.

Procedures

Eligible individuals who volunteered to participate were assigned to one of four groups based on the availability of classes and their preferred schedule and training location (see Table 2). Classes were held in the community at Dad’s Garage Theatre or Emory’s Integrated Memory Care Clinic. Pilot testing began once we had 8 potential participants. To compensate for their time and any costs, participants received $75.00 for each class they attended and for participating in Time 3 data collection, which included the final survey and in-person focus group sessions. Participants completed paper and pencil surveys in person with members of the research team available on site to assist. The program evaluation data reported here come from structured survey questions. Analysis of the study’s qualitative data are forthcoming and will appear in a subsequent article.

Table 2.

Select Training Logistics and Details by Learning Group

	Learning Group
Section Details	1	2	3	4
Time of Year	Fall	Winter	Winter	Spring
Day and Time	Wednesday Afternoons	Monday Evenings	Monday Evenings	Wednesday Evenings
Training Location	Dad’s Garage	Emory’s IMCC	Dad’s Garage	Dad’s Garage
Group Size	n=5	n=12	n=12	n=14
n (%)
# Completed
2 Classes	1 (20)	-	-	-
3 Classes	1 (20)	4 (33)	3 (25)	3 (22)
4 Classes	3 (60)	8 (67)	9 (75)	11 (78)
Completed T1, T2, T3 surveys	4 (80)	12 (100)	10 (83)	14 (100)
Focus Group Participation	4 (80)	12 (100)	10 (83)	14 (100)

Note: IMC = Integrated Memory Care Clinic.

As outlined in the Georgia State University’s (GSU) Institutional Review Board-approved protocol (H23627), prior to participant enrollment and study participation, researchers reviewed informed consent documents, which outlined the study’s procedures, risks, benefits, and voluntary nature of their participation. After obtaining written consent, we used assent procedures throughout the study. Electronic and paper records were deidentified and securely stored and accessible only to designated project staff.

Quantitative Measures

At baseline, we collected care partners’ demographic characteristics and key information pertaining to their care role, history, experiences, and about the person living with dementia. At all three timepoints, we also used several well-established measures to assess the program’s preliminary efficacy. We measured caregiver mastery, the primary outcome, using the subscale, Caregiving Competence (α=.74), from Pearlin Mastery, Loss, and Competence (Pearlin et al., 1990). This 4-item subscale uses a 4-point Likert scale with higher scores indicating greater competence and self-confidence. Widely used in caregiving research, the Pearlin scale shows good internal consistency (Chan et al., 2018; Hepburn et al., 2022; Pearlin et al., 1990). We examined the intervention’s efficacy on exploratory outcomes, including care partners’: perceived stress, using the Perceived Stress Scale, a 14-item Likert-type questionnaire (Cohen et al., 1983); depression, using the Center for Epidemiological Studies-Depression Scale (Radloff, 1977); anxiety, evaluated using the State-Trait Anxiety Inventory (Spielberger & Sydeman, 1994), a 20-item measure with 4-point Likert scale; caregiver burden, assessed with the Zarit Burden Inventory (Zarit et al., 1980), a 22-item scale with a 4-point Likert scale; and perceived changes in care recipients’ behavioral expressions using the Revised Memory and Behavioral Problems Checklist (Teri et al., 1992), which is comprised of 24 items with 4-point Likert scale responses. We also asked participants to assess care recipients’ quality of life using the Quality of Life-Alzheimer’s Disease Scale (Thorgrimsen et al., 2003). This scale consists of 13 items with a 4-point Likert scale. All exploratory measures have high internal consistency (Cronbach’s alpha >0.81) and a history of successful use among informal dementia care partners.

Evaluation Data

We collected evaluation data through surveys administered after participants completed the training (T2) and at month three (T3). T2 surveys administered immediately upon completion (Month 1) asked participants to rate the overall quality of the training program using a 5-item scale ranging from “very poor” to “very good.” We also invited comments or suggestions about the program or instructors. T3 surveys contained an item asking participants to report, on a 5-item scale ranging from “never” to “nearly always,” how frequently they use the skills taught in the program with their person. If participants reported using skills, they were asked to list any specific approaches used and how they were helpful.

Analyses

Survey responses were entered, stored, and managed using REDCap secure electronic data capture tools hosted at GSU (Harris et al., 2019; Harris et al., 2009). Data were imported into IBM SPSS for Windows, version 29.0 (IBM Corp, Armonk, NY) for analysis. After data cleaning and re-coding reverse variables, we calculated descriptives for demographic variables and instrument scores at baseline, Month 1 (program completion), and Month 3, as well as for all program evaluation questions. We conducted dependent-sample t-tests with Cohen’s d (Cohen, 1988) to assess pre-post differences on caregiver mastery, the primary outcome, and all exploratory outcomes. Cohen’s d can detect “small” (d=0.2), “moderate” (d=0.5), and “large” (d=0.8) effects, which facilitates interpret the intervention’s efficacy on study measures.

Results

Participants

Our needed sample size was 40 participants to complete the training. Forty-three care partners of people living with moderate dementia to participate enrolled in the study and began the training. This increased enrollment number had prior approval from GSU’s IRB. Of those who enrolled, 98% (n=42) attended at least three of four classes and 90% (n=40) completed data collection at all three time points. Of the three individuals who did not complete, one participant experienced an unanticipated change in employment after attending the second class and two had family crises unrelated to the training that prevented them from completing the final survey and attending the focus group. Due to low attrition, we did not evaluate differences between completers and non-completers.

Select baseline characteristics of those who completed the study appear in Table 3. Like the national profile of informal dementia care partners (Alzheimer’s Association, 2025), our participants ranged in age with most identifying as female and white, caring for a parent or parent-in-law, and living with the person. Our sample was, however, more highly educated with nearly everyone having completed college or obtaining an advanced degree, and did not reflect the racial and ethnic composition of caregivers found nationwide. Black care partners represented 18% of our sample, 2% were Asian, and none identified as Hispanic or Latino.

Table 3.

Select Sociodemographic Characteristics and Care Context of Participants who Completed the Trial (N=40)

Sociodemographic Characteristic	M [min-max]/ N (%)
Age - care partner	59.87 [34–81]
Gender Identity
Female	35 (87.5%)
Male	5 (12.5%)
Racial Identity
African American/Black	7 (17.5%)
White	31 (77.5%)
Asian	1 (2.5%)
Other	1 (2.5%)
Education
Some College	5 (12.5%)
College Degree	16 (40.0%)
Post Graduate Degree	19 (47.5%)
Employed
Yes	20 (50%)
Care Context
Person Living with Dementia
Spouse	13 (32.5%)
Parent (includes parent-in-law)	21 (52.5%)
Other (friend, sibling, grandparent)	6 (15.0%)
Age- Person Living with Dementia	79.85 [59–96]
Cohabits with Person	28 (70.0%)
Months Spent in Care Role	49.44 [2–3601]
Perceived Help from Others
Some or a great deal of help	24 (60.0%)
Little to no help	15 (37.5%)
No response	1 (2.5%)

Feasibility and Acceptability

During our pilot study’s 5-month recruitment window, 60 individuals contacted our team and were screened. Of those, 48 caregivers met study criteria; five were unable to attend the training program because of scheduling or an unexpected health crisis. Nevertheless, our enrolment of 43 care partners during this time provides evidence of feasibility and interest in improv-based training. Our ability to retain nearly all participants to training completion (n=42) speaks to both the feasibility and acceptability of the program as do evaluation data.

Ninety-eight percent 98% of those who completed the T2 surveys said that the overall quality of the training program was “good” or “very good,” the two highest ratings on the 5-point scale. Over three quarters (n=33) provided a response when invited to share any comments or suggestions about the training program or instructors. All responses were positive and identified strengths of the program or instructors and/or offered constructive suggestions for improving the program. Ten percent of participant comments noted the training team as a strength; 45.5% highlighted the training itself, especially the interactive mode of instruction, class content, and in-person and group format. Over one-third (36.4%) of responses included suggestions for expanding the training by providing additional information or materials (e.g., more emphasis on application, the addition of video resources) or by making sessions longer or having an additional class or classes.

Preliminary Efficacy

Our quantitative analysis, reported in Table 4, suggests that “Improving Care through Improv” influenced the primary outcome of caregiver mastery as well as select exploratory outcomes, namely caregivers’ stress, depression, and reaction to care recipient’s behaviors. Caregiver mastery showed improvement between baseline (T1) and course completion (T2/Month 1) and from baseline (T1) to follow up (T3/Month 3). Improvements from baseline and completion and between completion and follow-up were not statistically significant. However, participants’ mean competence score increased from 11.10 to 11.93, a change of 0.83 points, on average, between baseline and follow-up. This finding was statistically significant (p < .01) and yielded a moderate effect size (Cohen’s d = −0.463).

Table 4.

Primary and Exploratory Outcomes Assessed at Baseline (T1), Month 1/Training Completion (T2), Month 3/Follow up (T3)

	M (SD)
Measure	Time 1 (Baseline) n=43	Time 2 (Month 1) n=42	Time 3 (Month 3) n=40	p Value (T1-T2)	p Value (T1-T3)	p Value (T2-T3)	Cohen's d (T1-T2)	Cohen's d (T1-T3)	Cohen's d (T2-T3)
Mastery	11.10 (2.31)	11.63 (2.07)	11.93 (2.06)	0.053	0.006	0.244	−0.316	−0.463	−0.187
Perceived Stress	26.25 (8.43)	26.65 (9.45)	23.10 (8.16)	0.609	0.001	0.001	−0.081	0.551	0.555
Depression	14.83 (9.30)	14.78 (9.88)	12.98 (9.40)	0.957	0.037	0.104	0.008	0.342	0.263
Anxiety	36.15 (10.27)	35.38 (11.09)	34.85 (11.62)	0.633	0.384	0.730	0.076	0.139	0.055
Burden	61.95 (16.32)	61.63 (15.19)	59.68 (16.43)	0.787	0.135	0.121	0.043	0.241	0.250
Behavioral Expressions
Frequency	37.20 (14.21)	36.55 (11.99)	38.05 (13.78)	0.597	0.550	0.202	0.084	−0.095	−0.205
Reaction	20.13 (13.27)	19.70 (12.21)	16.35 (12.05)	0.945	0.035	0.013	0.011	0.349	0.413
Quality of Life – Person Living with Dementia	27.85 (5.15)	27.88 (4.20)	27.63 (5.16)	0.967	0.680	0.605	−0.007	0.066	0.082

There were no statistically significant changes in exploratory measures between baseline and upon program completion. However, perceived stress decreased from 26.25 at baseline to 23.10 follow up (p < .001) with a moderate effect size (Cohen’s d = −0.551). A similar pattern was observed with care partners’ depression. The mean depression score improved from 14.83 at baseline to 12.98 at follow-up (p < .05) with a small-to-moderate effect size (Cohen’s d = 0.342). There was a decrease in care partners’ reactions to their care recipients’ behavioral expressions between baseline and follow-up. The mean score changed from 20.13 to 16.35 (p < .05) with a small-to-moderate effect size (Cohen’s d = 0.349). Although less impacted, there was no significant change in care partner-reported frequency of behavioral expressions. Anxiety and burden decreased, but effect sizes were small and not significant. The training did not influence participant evaluations of the quality of life of the person with dementia.

When invited to name any specific strategies being used because of the training, 83% of care partners who completed the training and T2 surveys (n=35/42) identified at least one approach. Of those, the majority (71.4%) identified “Yes, and…” and/or referenced related strategies such as not arguing, being flexible or adaptable, and trying to be in the moment. In follow-up surveys completed at Month 3, two-thirds of participants said they were using the skills, “quite frequently” or “nearly always.” The remaining third said, “sometimes.” When prompted to list specific strategies and how they are helpful, seventy-five percent of care partners responded. Once again, the majority, in this instance, two-thirds listed “Yes, and…” or commented on the successful use of validation, accepting their person’s reality, not arguing, and making their partner look good.

Discussion

As the number of people living with dementia increases, so does the need for effective evidence-based programing to equip the millions of people who are or will be, providing the bulk of care. “Improving Care through Improv” responds to this urgent public health need. Results point to the program’s feasibility, acceptability, and preliminary efficacy for enhancing mastery and aspects of emotional well-being. Findings have implications for research and practice.

Our pilot test affirms findings from our in-depth systematic observational of meaningful engagement and quality of life among assisted living residents with dementia. In this foundational study, we learned that care partners, including staff, external care providers, volunteers, and residents’ family and friends who had the greatest success engaging and interacting with people living with dementia used techniques that parallel improv (Kemp et al., 2025; Kemp et al., 2021; Kemp et al., 2024). By developing and testing training for informal care partners, we were better able to understand how and why using improv skills may be beneficial and lead to positive care partner and recipient outcomes.

Alongside evaluation data, successful recruitment and high rates of retention/training (98%) and overall study completion (90%) demonstrate feasibility and acceptability. These data indicate the demand for and interest in our program and our ability to overcome concerns related to recruiting and retaining dementia caregivers, especially in a pandemic/post-pandemic time. Although effective evidence-based psychoeducational programs exist, “Improving Care through Improv” may be perceived as more of an activity that is fun and engaging and therefore more widely appealing to care partners relative to traditional caregiver training programs.

According to participants, the peer group in-person format and active learning methods provided social benefits and engagement that further reinforced course content and skill development. With the potential to be more cost effective and to reach more people simultaneously, group training is apt to be more economical and efficient relative to one-on-one training (McDermut et al., 2001). Additionally, as our participants noted, group training provides opportunities for peer support and sharing (Leung et al., 2015).

Our work advances, in significant ways, the important research from the two known studies piloting improv training for caregivers (see Brunet et al., 2021; Howell et al., 2022). Our curriculum was informed by our qualitative research. Our study used social cognitive behavioral theory as a guiding framework and involved the assessment of caregiver mastery. We delivered group-based in-person improv training for dementia care partners involving 2-hour weekly sessions. Our program, however, consisted of fewer sessions, four as opposed to six weeks and used a threshold of 75% class attendance for training completion. These programmatic differences may have contributed to our higher retention rates. Yet, our high study completion rates are notable, especially given that we assessed outcomes at three rather than two timepoints. Caregiver interventions often have high participant dropout rates (Gilhooly et al., 2016) making our completion rates exceptionally encouraging.

Several of our exploratory outcome variables overlapped with previous pilot studies as did certain findings. Unlike Brunet and Howell’s teams, we observed statistically significant changes between baseline and follow-up rather than between baseline and program completion. These study results aligns with research on adult learners and caregiver education, which suggests the need for reflection and time to practice and apply skills (Cooke et al., 2001). This observation also speaks to the potential value of a booster session, which would meet certain participants’ desire for additional education and allow time for processing and application. Thus, while the training met or exceeded participants’ needs and expectations, there is room for improvement and growth.

Similar to the two existing improv studies, we measured depression, burden, and evaluated the care recipients’ neuropsychiatric symptoms and care partners’ associated distress. Our intervention showed preliminary efficacy for reducing depression but not burden as both other studies found. The program’s lack of success in this domain is, however, consistent with results from Hepburn and colleague’s (2022, p. 622) randomized controlled trial of Tele-Savvy. Characterizing the lack of success in reducing burden as “understandable,” Hepburn et al. point to Zarit et al. (1980) who found that caregiver burden was most strongly associated with frequency of family contact and other social support measures.

We found no significant change, either positive or negative, in care partner reports of care recipients’ behavioral and psychological symptoms. However, consistent with Brunet et al. (2021) and Howell et al. (2022), training reduced care partners’ negative responses, an encouraging finding that reinforces Social Cognitive Theory (Bandura, 1977, 1989, 2001) and the concept of mastery. Given that the person with dementia was not the intervention’s focus, changes in their behavior seem less likely. However, because the program is designed to equip care partners with new skills and ways of coping, ideally there will be accompanying reductions in negative responses (see also Hepburn et al., 2022). Further reinforcing this hypothesis, our program was associated with decreased self-perceived stress upon completion and at three-months. This stress measure, along with anxiety and a proxy measure evaluating the quality of life of the person living with dementia, were included in our study but not measured in the others. We found no significant change in anxiety or the care recipients’ quality of life.

Notably our study was based on a significantly larger and sufficiently powered sample, yielded a higher retention rate, and included an additional evaluation time point three months post-baseline to understand the durability of the intervention and its ability to maintain benefits. We are encouraged by participants’ evaluation data, the positive response to the program, descriptions of application, examples of what they learned, and reports of having “new ways to handle situations.”

Strengths, Limitations, and Future Directions

Our study has several strengths and limitations to inform future work that warrant consideration. To begin, use of quantitative and qualitative approaches enhance our trial. While not presented here, analysis of our qualitative data, will allow for the triangulation of data on feasibility, acceptability, and preliminary efficacy. In future, however, additional qualitative data analysis and collection, especially in the form of additional observations and periodic open-response questions and process evaluation, would enhance understanding of participants’ application efforts and experiences, including accompanying success, failures, challenges, and responses from their person. McLoughlin’s (2022) systematic review of group-based interventions for carers of people living with dementia recommends the use of such methods to understand participants’ experiences considering their often highly heterogeneous care situations.

Our study pilot tested the training program among informal care partners using a sufficiently powered sample. Yet, with our convenience as opposed to maximum variation sampling methods, our participants were highly educated, presumably with access to greater resources than the typical care partner. Presumably, these resources allowed them to step away from other responsibilities, including paid employment, school, or care work, to participate in an in-person training program. Future recruitment efforts should seek greater heterogeneity in terms of educational and socioeconomic status as well as race, ethnicity, gender, sexual orientation, and other important characteristics apt to influence care partners’ experiences and hence the program’s feasibility, acceptability, and efficacy. Moving forward, it will be critical to understand the barriers to participation and learn how, if at all, the intervention could be adapted to broaden care partner access.

Neither our study nor those of Burnet and Howell’s teams involved control groups, which limits our ability to compare groups. Future work should use control conditions to facilitate statistical comparison and enhance understanding of the intervention, including its components and influence on participants. Use of control group(s) such as waitlist or attention control groups would address potential concerns about placebo effects or reduce bias.

Unlike the two known existing pilot tests of improv training for family members, we assessed caregiver mastery, our primary outcome, to measure increases in competence and confidence. Although insufficiently powered to conduct mediation analysis, a larger sufficiently powered trial should test the relationship between mastery and caregiver emotional well-being. Doing so would advance understanding of the mechanism of action (i.e. mediator) and allow us to understand why changes in caregiver wellbeing might occur following the training. We hypothesize that caregiver mastery is the mechanism of action that mediates the relationship between “Improving Care through Improv” and care partner outcomes.

Improv techniques, as we note elsewhere, may not work in every scenario, including in extreme instances of aggression” but rather, “offer a set of strategies” and “under the right circumstances, can be helpful” (Kemp et al., 2024, p. 307). It is critical to assess, understand, and acknowledge and the strengths, limitations, and appropriate uses of improv skills. Understanding how approaches work or do not work, for example, with people with different types of dementia, such as those with Frontotemporal Disorders, or with specific types of behavioral changes, is an important future consideration.

Conclusion

Our study contributes to the growing body of interdisciplinary work that demonstrates the benefits of improv training, including among dementia care partners. Our encouraging findings lay the foundation for further work. Part of improv’s potential lies in its power of connection, meeting people where they are, in the moment, and on their terms. As Metha et al. (2021, p. 265) notes, “‘Yes, and’ doesn’t imply indiscriminate acquiescence or capitulation . . . but rather [it means saying] ‘yes’ to understanding [another’s] view ‘and’ ways to connect as we find common ground.” “Improving Care through Improv” holds promise in finding common ground in the dementia care context. In so doing, it has the potential for innovating dementia care education and enhancing dementia care competencies among different types of care partners in a variety of care settings.

Data Availability:

Deidentified quantitative data from this study will be made available to qualified researchers upon reasonable request through a data sharing agreement that is consistent with NIH data sharing policies, as well as policies and practices established by the Roybal Center for Dementia Caregiving Mastery at Emory University. To protect participants, qualitative data are not available.