Moderators of Adherence Within a Nursing Home Pragmatic Trial of Personalized Music

Abstract

Background and Objectives

To describe factors affecting nursing home adherence in an embedded pragmatic randomized controlled trial of a personalized music intervention to manage agitation in residents living with dementia.

Research Design and Methods

Semistructured qualitative interviews with 4 corporate leaders and 27 nursing home staff. We used thematic analysis to develop a codebook and map findings to potential adherence moderators in the Conceptual Framework for Implementation Fidelity (CFIF): recruitment, participant responsiveness, strategies to facilitate implementation, intervention complexity, quality of delivery, and context.

Results

Recruitment: Corporate leaders noted research participation compensation did support corporate implementation. Resident turnover frequently occurred and led to delays in implementation due to the need to personalize music to each resident. Participant responsiveness (dose received): Interviewees noted the intervention improved dementia behaviors and enhanced engagement with exceptions. Strategies to facilitate implementation: Interviewees voiced differing views on whether nurses or activities staff should lead implementation and how to pilot the program. Intervention complexity: Interviewees described complexities including tailoring delivery to each resident, updating music, and sustainability concerns. Quality of delivery: Champions discussed protocol deviations; for example, providing residents with music that was not personalized. Context: Interviewees noted how the SARS-CoV-2 pandemic exacerbated turnover, but the intervention did engage residents during staffing shortages.

Discussion and Implications

Interviewees described contextual barriers associated with pragmatic implementation of a personalized music intervention. While they also described facilitators, our findings highlight limits to pragmatic delivery in nursing homes with chronic under-resourcing and staffing, exacerbated by the pandemic.

Background and Objectives

It is challenging for nursing home staff to engage in embedded pragmatic clinical trials (ePCTs), trials in which they, rather than researchers, are responsible for implementing interventions (Resnick et al., 2022). This is in part because nursing home staff care for residents with chronic conditions and high needs while mitigating resource constraints and short staffing (National Academies of Sciences, 2022). The SARS-CoV-2 pandemic intensified nursing home workforce issues and added workload, new regulatory requirements, and challenges related to caring for a high-risk population. This makes nursing home ePCT implementation fidelity a growing concern. These trials are not designed for controlled settings where protocol adherence is assured (Vernooij-Dassen & Moniz-Cook, 2014).

The ePCT of Music & Memory (Brown University, 2021), a personalized music intervention for managing agitation in nursing home residents living with dementia, provides an opportunity to examine factors affecting adherence in a trial that occurred during the pandemic. The overall study consisted of two parallel ePCTs, each with 54 nursing homes randomized to provide the intervention (N = 27) or serve as usual care controls (N = 27). This Trial adds to an existing body of evidence surrounding the effects of personalized music on agitation in nursing home residents with ADRD. While two large quasi-experimental studies found effects of personalized music on agitated behaviors in nursing home residents with ADRD in the past 10 years (Bakerjian et al., 2020; Thomas et al., 2017), these findings were not replicated in the small (59 residents, 10 NHs) randomized controlled trial conducted before this trial (Kwak et al., 2020). This is the first large RCT of personalized music for agitation in the United States. The first trial finished before the SARS-CoV-2 pandemic; the second trial, after it began. Results from the prepandemic trial have been published elsewhere (Davoodi et al., 2023; McCreedy et al., 2019, 2021, 2022; Olson et al., 2022; Zediker et al., 2023; Zhang et al., 2023) and found that personalized music was more engaging than less personalized music, but that personalization was time-consuming and mostly completed by activities staff, rather than the nursing staff who interact most with residents and may be best positioned to use music with them regularly. In Trial 1, there was a positive effect of the intervention on verbal agitated behaviors but not physical agitated behaviors (Sisti et al., 2023). There was a near-significant effect on antipsychotic use, but no effect on staff-reported behaviors, depressive symptoms, or other medication use (McCreedy et al., 2022). However, the personalization process took staff an average of 2.5 hr per resident in Trial 1 (Davoodi et al., 2023), and Trial 2 was necessary to see if a more streamlined approach to personalization could still be effective (McCreedy et al., 2021). In the second trial, research staff identified and loaded residents’ music on music players and encouraged nursing staff to use the music with residents at times when agitated behaviors were likely. Despite implementation challenges during this second trial, including staffing shortages and new infection control precautions, the program was associated with improved resident agitation and mood (McCreedy et al., 2024).

In this paper, we use the Conceptual Framework for Implementation Fidelity (CFIF) to describe staff perceptions of factors affecting adherence in the second trial. CFIF was specifically designed to assist researchers in evaluating complex interventions and has been used to characterize factors affecting adherence in nursing home ePCTs. Figure 1 presents the CFIF framework and how potential modifying factors to adherence may shape the implementation fidelity of an intervention. We mapped data from semistructured, qualitative interviews with nursing home leaders and staff engaged in the study to CFIF’s adherence moderating factors (Carroll et al., 2007; Hasson, 2010). We anticipate our findings will be important to other pragmatic trialists navigating the nursing home environment since the pandemic onset.

Figure 1.

Modified conceptual framework for implementation fidelity (Hasson, 2010)

Research Design and Methods

Setting and Population Studied

Brown University conducted Music & MEmory: A Pragmatic TRial for Nursing Home Residents With ALzheimer’s Disease (METRIcAL) in partnership with four U.S. nursing home corporations (ClinicalTrials.gov Identifier: NCT03821844). Research activities paused between March 2020 and May 2021 to accommodate nursing home staffing constraints during the early days of the COVID-19 pandemic, prompting the research team to modify the study design to include two separate parallel trials, each enrolling 54 facilities (27 treatment, 27 control). One trial was conducted from June 2019 to February 2020 and has been published elsewhere (McCreedy et al., 2022), and the other, the focus of this paper, was conducted from May 2021 to September 2022. We interviewed the corporate- and site-level staff responsible for leading implementation of the second trial in their corporation or nursing home.

Interviewees

Corporate interviewees were the “corporate leader,” serving as point of contact with the research team and sites and training participating sites’ staff on implementing the study protocol. The four participating corporations received payments to defray participation costs, including data transfer and assigning a corporate trainer to lead the study site implementation. No incentives were provided for nursing home staff to administer the music to residents. These incentives did not affect the trial’s pragmatic design from the nursing home perspective. The four corporate leaders were the only interview participants whose participation spanned the first and second ePCTs.

Nursing home interviewees were the staff member designated “site champion” and responsible for oversight of their nursing home’s implementation of the intervention. At the trial’s start, researchers asked nursing home corporate leaders to select champions, preferably a nursing staff member (e.g., director of nursing, assistant director of nursing, or unit manager) or the staff member best positioned to lead local implementation efforts. Champions served as liaisons with the research team and their colleagues, including ensuring implementation in their facilities.

Intervention

The individualized music intervention and protocol have been detailed elsewhere (McCreedy et al., 2021). Researchers loaded residents’ preferred music on an MP3 player based on family, staff, or resident-reported preferences (when available). When residents’ preferred music was not known, research staff loaded music that was popular when the resident was between the ages of 16 and 26. Researchers labeled the MP3 players with residents’ initials and provided the devices, headphones, small speakers (for use with residents who did not tolerate headphones), and chargers to site champions. The protocol suggested using the music with participating residents at times of day when target behaviors were likely.

Analytical Framework

We defined the CFIF domains (Hasson, 2010; Figure 1) as follows: recruitment, as factors driving corporate participation and how site champions recruited eligible residents; participant responsiveness, as resident and staff reactions to the personalized music intervention; strategies to facilitate implementation, as training and efforts to tailor and sustain delivery; intervention complexity, as how interviewees described the program and its complexity; quality of delivery, as how the intervention’s delivery adhered to the protocol, including music device distribution to residents, and staff ensuring ongoing use; and context, as organizational culture or co-occurring and historical events affecting implementation, such as the pandemic.

Approach and Analysis

We developed semistructured interview guides for corporate leaders and site champions. The corporate leaders’ interview guide included 11 questions and was intended to be completed in 30 min; the champion interview guide contained seven questions and lasted 5–10 min (see Supplementary Materials). The research team designed the semistructured interviews to evaluate processes and opportunities for improvement within the trial. We mapped findings to the CFIF after an initial review of interview transcripts as the findings directly related to the CFIF moderators of adherence.

The project director (A. R.) scheduled interviews and a research assistant (G. W.) conducted them. Research assistants (G. W. and two others) called each corporate leader and site champion to complete the interview or schedule an interview time. Interviews were conducted close to the end of the last resident’s follow-up window, audio-recorded with verbal permission, and then transcribed. Four of the authors (J. W., G. W., R. B., and E. M.) coded the transcripts using thematic analysis (Braun & Clarke, 2006) to identify the most relevant themes pertaining to the trial’s implementation. We developed our codebook using deductive coding, met as a team to refine the codebook and gain consensus on coded excerpts, and mapped the codes to the corresponding CFIF moderators. At least two authors coded each transcript. All four authors met to reach consensus in mapping codes to the CFIF domains. We then used nVivo 12.0 Plus to generate analytic reports.

We provide national comparisons to intervention nursing homes in METRIcALas nursing homes in this study were recruited nonrandomly. Descriptive statistics allow us to examine important differences in our sample that could shape implementation and qualitative responses. We examined publicly available data from LTCFocus (2019) including: residents’ race (black, white); gender (male, female); % of residents with an Alzheimer’s/dementia diagnosis; resident ADL score; registered nurse, licensed professional nurse, and certified nursing assistant hours per resident day; and total bed count. We also compiled Medicare.gov Star Quality Ratings, which range from one to five, with higher scores representing better quality (Centers for Medicare & Medicaid Services, 2019). We used t tests and 95% confidence intervals to test for differences between intervention and all U.S. nursing homes.

The Brown University Institutional Review Board (IRB) approved the conduct of the trial with a waiver of informed consent, as a minimal-risk trial (protocol number: 1705001793). The Brown University Institutional Review Board does not consider the qualitative interviews to be human subjects research, because the participants were speaking in their professional capacity and did not provide any information about themselves. Therefore, this project did not require IRB approval.

Results

Four corporations participated in the trial: two not-for-profit; two for-profit. They are located in the Midwest (2); mid-Atlantic (1); and the South (1). The 27 nursing homes randomized to the intervention were statistically similar to national averages in nursing staffing levels and total beds but had higher occupancy rates (91.7% vs. 84.0% nationally) and lower Centers for Medicare & Medicaid Services (CMS) Overall and Long-Stay Quality 5 Star Ratings (Overall: 2.6 vs. 3.1 nationally; Quality: 2.8 vs. 3.6 nationally). Residents in the intervention nursing homes were similar to national statistics in median age (66.1 years vs. 66.3 years nationally), residents with Alzheimer’s disease or related dementia diagnoses (48.7% vs. 50.6% nationally), and ADL scores (16.4 vs. 16.7 nationally). However, they had a significantly higher median percentage of Black residents (19.5% vs. 4.3% nationally).

We interviewed all corporate leaders (N = 4) and site champions (N = 27) from four corporations and 27 participating nursing homes. Corporate leader interviews averaged 29.9 min (range, 21–35 min); champion interviews, 8.3 min (range, 3–21 min). Corporate leaders’ held positions involving working with facilities on staff education, dementia care, and quality. Despite requesting site champions be from nursing staff, only six (22.2%) were nursing staff; 14 (51.8%) were activities staff; and 7 (25.9%) were from another department like administration, or memory care (Table 1). Of the 431 residents in the music condition, 294 (68.2%) had some genre/song/artist preference information provided by the nursing home which was then used by the research team to personalize music on the resident’s device. Of the 190 residents in the music condition who had any play data, 150 (78.9%) had some genre/song/artist preference information provided by the nursing home.

Table 1.

Baseline Descriptive Characteristics, 2019

Characteristic	Intervention nursing homes (N = 27)	U.S. nursing homes (N = 15,322)	p Value
Nursing home
Licensed beds, median number (IQR)	100 (79–135)	100 (66–128)	.87
Occupancy, median % (IQR)	91.7 (84.3–94.8)	84.0 (71.8–91.7)	.003*
5 star rating, mean (SD)a
Overall	2.6 (1.3)	3.1 (1.4)	.04*
Long-stay quality measure rating	2.8 (1.3)	3.6 (1.3)	.04*
Medicaid %, median (IQR)	69.3 (45.7–80.6)	65.2 (50.0–77.1)	.30
Staffing, hours/resident/day (IQR)
Registered nurses	0.4 (0.2–0.5)	0.4 (0.2–0.6)	.44
Licensed practical nurses	0.8 (0.5–0.9)	0.8 (0.6–1.0)	.33
Certified nursing assistants	2.0 (1.6–2.2)	2.1 (1.8–2.5)	.01*

Resident
Female, median % (IQR)	66.1 (54.1–75.0)	66.4 (58.0–73.3)	.67
White, median % (IQR)	87.5 (53.1–99.1)	87.3 (67.8–96.5)	.39
Black, median % (IQR)	19.5 (0.0–46.0)	4.7 (0.0–28.2)	.04*
Alzheimer’s or dementia diagnosis, median % (IQR)	48.7 (43.1–55.4)	50.6 (40.6–60.4)	.60
Activities of daily living, median score 0–28 (IQR)b	16.4 (14.5–17.5)	16.7 (15–18.1)	.57

Notes: IQR = interquartile range; SD = standard deviation; * = p value ≤ .05.

Data derived from CMS Medicare.gov Data Archive and LTCFocus Public Use Files, 2019.

^a5-Star Quality Ratings are determined by CMS by assessing the facility’s relative performance on certain quality metrics, staffing, and health inspections. Scores range from 0 to 5 with higher scores indicating better overall performance (Centers for Medicare & Medicaid Services, 2023).

^bThis score measures an individual’s independence on 7 ADLs—bed mobility, transfer, mobility, dressing, eating, toilet use, and personal hygiene. Each ADL is scored from 0 to 4, with 0 indicating total independence in that ADL and 4 indicating total dependence in that ADL, or the activity didn’t occur. The ADL score range is from 0 to 28, where 0 indicates completely independent and 28 completely dependent (Morris et al., 1999).

Table 2 presents the qualitative findings, including the operational definitions, primary themes, and summaries for each identified domain.

Table 2.

Interview Themes, by Conceptual Framework for Implementation Fidelity Domain

Operational definition	Themes	Summary
Recruitment: What recruitment procedures were used to attract individuals to the intervention? What constituted barriers to maintaining involvement of individuals?
What factored into corporate decisions to participate? How did site champions identify residents to participate?	Corporate leaders deciding research involvement Incentives adequate, valuable Desiring to benefit residents with dementia Replacing residents in the intervention	While corporate leaders appreciated monetary incentives for research participation, opinions varied on how to use those funds and champions were largely unaware of them

Participant responsiveness (dose received): How were the participants engaged with the intervention? How satisfied were the participants with the intervention? How did the participants perceive the outcomes and relevance of the intervention?
What resident reactions were observed by site champions? When/how did staff members utilize the devices? Did the intervention help address dementia-related agitation?	Staff enjoying & bonding Residents reminiscing, discussing music Varying resident responses Helping staff in addressing behaviors (anxiety, aggression, wandering, anxiousness, agitation, anger, combativeness, etc.) Some residents disliking music choice, headphones, throwing devices Intervention timing for behaviors Not as effective for residents with advanced dementia, increasing confusion Residents smiling, moving, singing along with music	Corporate leaders and champions described residents’ reactions (e.g., its effect on resident outcomes and on resident-staff relationships) and staff perceptions of the impact on workload

Strategies to facilitate implementation: What strategies were used to support implementation? How were these strategies perceived by staff involved in the project?
What staff communication strategies facilitated program implementation? What changes did staff recommend to increase program sustainability and flexibility?	Educating new staff Checking in with staff Preloading iPods with music Appropriate role of site champion Implementing and communicating facility-wide	Corporate leaders and champions shared strategies to improve implementation, including which department should lead, piloting with a few residents before expanding, and ensuring facility-wide adoption

Intervention complexity: How complex is the intervention? How well is it described?
What components of Music & Memory affected staff ease of use or effort?	Charging, storing devices Using headphones versus speakers Updating and loading music Integrating into workflow Mailing devices Time consuming for staff Sustainability	Corporate leaders and champions described complexities, from tailoring the music’s delivery to each resident to mailing the MP3 players to researchers for data extraction

Quality of delivery: How was the quality of delivering the intervention’s components? Was the delivery process appropriate in achieving what was intended? Did staff receive appropriate support and resources to fully implement the intervention according to protocol?
What were the experiences of the site champions in using the music devices? What technological issues did they experience during the program?	Adhering to the protocol (and lack of) Delaying the start date Device functionality, charging issues	Champions discussed how they did not always adhere to the protocol, some, for example, by providing residents with music playlists that were not personalized

Context: What factors at political, economical, organizational, and work group levels affected the implementation?
How did the COVID-19 pandemic impact overall implementation? What facility-level characteristics were important for implementation success?	COVID-19 pandemic causing issues with keeping track of devices Staffing challenges and site champion turnover	Corporate leaders and champions discussed how the pandemic affected implementation, for example by increasing their desire for implementation flexibility and resulting in champion turnover

Note: COVID-19 = coronavirus disease 2019.

Recruitment: Factors Related to Corporate Participation and Ascertaining Residents’ Eligibility

In themes mapped to recruitment, corporate leaders discussed corporate-level decision-making to participate in the study, while champions spoke about resident-level recruitment.

Corporate leaders described their corporations’ decision-making process for engaging in research typically including members of their C-suite, operations, and clinical leaders. They discussed the intervention’s alignment with corporate goals to reduce off-label medication use and to implement nondrug interventions to improve dementia care. Corporate leaders expressed appreciation for financial incentive payments in offsetting operational costs, but they did not consider them a key factor in their corporate participation decision. They described the intervention’s potential benefits for residents as their primary motivation.

I think it definitely encourages us as an organization to participate, but I don’t think, and particularly with the Music & Memory program, that it actually drove that main decision to participate. I think we just wanted to be part of that research, let’s see where it goes, let’s see if it does benefit our residents, and let’s implement it if so. (S3, Corporate Leader)

Corporate leaders supported recognizing individual nursing home staff but did not use the payments for staff incentives.

I don’t know if maybe [financial incentives] should have been looked at more or done more… I was just so focused on giving our residents something that was non-pharm, the money didn’t matter to me, so I didn’t have that as a focal point for me. But to have [nursing home staff] understand that they were going to be compensated for their efforts, I don’t know if that would’ve made a difference. But there, again, I guess I just don’t even know if they knew or they know about it. (S4, Corporate leader)

At the facility level, site champions described challenges mainly with recruitment after residents disenrolled (e.g., discharged, transferred, or died). Some did not realize they should contact the research team to create a personalized playlist for newly enrolled residents. Instead, they described deviating from the research protocol by giving personalized MP3 players to other residents they perceived it might benefit.

[We] had 15, but [a] 16th was sent to us [...] So the extra one, that’s the one [the activities director has] been having residents share, and I think they’re using the same music. (S3N9, Site Champion)

Participant Responsiveness: Resident and Staff Reactions to the Intervention and Its Delivery

In themes mapped to the participant responsiveness domain, corporate leaders and site champions described residents’ reactions (e.g., its effect on resident outcomes and resident–staff relationships) and staff perceptions of the impact on workload.

Although site champions were asked to select residents who experienced behaviors, they could also use the music with participating residents for pleasure. Some described how use varied from resident to resident, with one commenting that use was more frequent for those residents experiencing behaviors than for those whose use of the music was more for quality of life.

If somebody was having an out of character response, then that’s what it was used for. But if a person was just passively sitting in their room and not very vocal and it was more of a quality of life thing, it was less likely to be used. (S4N3, Site Champion)

While their descriptions of residents’ responses to the intervention varied, site champions typically reported that most residents enjoyed the music and seemed to experience happiness while listening to their playlists. Commonly reported resident reactions included changes in expression or smiling, moving or swaying to the music, dancing, humming, or singing.

[To] see those residents that appear to be unreachable enjoying the music, maybe seeing a tear come down their cheek…a cheerful tear – it just does something [for us] that actually see that. When they’re pretty much nonverbal and they don’t move and they sit in that chair and you put that music on, and then you suddenly see their foot or their knees swaying back and forth, they’re a little more content, relaxed. You kind of see an improvement. (S3, Corporate leader)

Site champions described using the music to engage residents and to address a variety of behaviors, ranging from wandering and depression to anxiousness, agitation, anger, and resistance to care activities. Some noted the timing of the intervention influenced the extent to which staff could redirect residents’ behaviors.

[One] of our ladies who will start pacing in the evenings, she’s a big sundowner. And if they don’t get her right away when they first start seeing [her] pacing… then the behaviors really start to show out. If you get her beforehand, she’s fine sitting there and it’ll calm her down. She sits there and listens to the music, and sometimes she’ll sing along. (S1N2, Site Champion)

…it’s been really good, really good. Because he gets kind of antsy and otherwise he’s playing in the ice machine and doing other stuff. But as long as he is listening to his music and he’s singing along to it, we want to keep it there. (S2N1, Site Champion)

While site champions generally characterized residents’ responses as positive, they also noted that some did not like the type of music or were confused by either the music or the MP3 players themselves. They also described how staff used the music more with participating residents when they observed positive responses. People who disliked wearing the headphones sometimes took them off or threw the MP3 players, and some residents with more advanced dementia either became confused or agitated or had no discernable reaction.

I know we had one lady who early on it was very beneficial and she was even kind of singing her words even when she wasn’t wearing the headphones. And just over the course of time, the progression has caused her to, “Get that off me. What is that? Get away from me,” that type of thing. (S4N3, Site Champion)

We do have [someone with] advanced dementia that does not really react at all. (S2N4, Site Champion)

Strategies to Facilitate Implementation: Training and Efforts to Adapt and Sustain Delivery

In themes mapped to the strategies to facilitate implementation domain, corporate leaders and site champions described site-level leadership of implementation and generally agreed that facility-wide implementation involving multiple departments was ideal.

Many corporate leaders emphasized the importance of site-specific planning before implementation as a facilitator of success.

[It] is something that requires due diligence and planning and caring through and then evaluating what works, what doesn’t work, and some of our facilities lacked in those areas. (S4, Corporate leader)

I think it really is a valuable benefit to both the individual that’s receiving the individualized music, as well as a great intervention for staff. I do think that the more buy-in you have from everyone, the easier it’s able to access. And having a good plan for how you are going to get all the music uploaded onto the iPods is a critical piece for this to be successful. (S2, Corporate Leader)

Site champions highlighted the need for more staff involvement, typically from nursing and activities, to mitigate capacity constraints although perspectives differed on which was most appropriate. Some suggested whoever had the most resident interaction should handle music delivery. Interview participants emphasized integrating the intervention across staff and departments and the importance of communication with staff via educational materials and direct communications. Many corporate leaders and site champions suggested facility-wide training and implementation to promote staff buy-in and long-term use of the program.

I think facility wise [it] would be to have more people helping with the program. That would’ve been more beneficial. (S4N4, Site Champion)

A part of it is just that, in my position. I’m busy with a lot of other things. I don’t know if it would maybe help if we had activities do it or something. (S2N5, Site Champion)

Some described checking in with staff to identify implementation barriers.

Chit chat with your staff and go and check on them [because they were] kind of struggling down the other hallways. Which, I’m activities on the dementia unit, so I’m always here. And so, I would grab those out for our residents. But, just [check] on your staff, see if there’s anything you can do to help them. (S4N1, Site Champion)

Site champions described their staff’s desire to be present when the music was used, to learn whether the intervention benefitted residents, and to build relationships.

Intervention Complexity: The Program, Its Complexity, and any Implementation Complexity

In themes mapped to the intervention complexity domain, site champions discussed issues related to tailoring music to residents and to the MP3 players and related accessories (e.g., headphones, chargers), including mailing the players to researchers for data extraction.

Corporate leaders discussed the value of partnering with the research team in the second trial to accomplish personalization and reduce staff burden, contrasting against the previous trial, where staff assumed this responsibility. One corporate leader described the more time-intensive process as valuable for staff-resident connections.

You guys helped tremendously in taking the time aspect away[...]. [The first trial,] they did everything. They loaded the iPods, we investigated, loaded the iPods and that was very, very time-consuming, but yet necessary. I absolutely appreciated that process because I found that made a better connection with the resident and the music that they had. (S4, Corporate Leader)

Champions spoke to the importance of personalizing the music, noting that participating residents benefitted from music they recognized and remembered.

A lot of times [...] we’ll have music playing on the TV and we’ll have residents in the living room, memory support, and they really enjoy that. So I did feel like personalized music, for the ones who are part of the program. [they] have really benefited from it. (S4N4, Site Champion)

However, a few noted that sometimes residents were startled by certain genres or the playlist order, and recommended single-genre or calming, instrumental music.

[It would be better] if we had just had those that was just soul music and then some that was just country or some that was just Christian, instead of the mixed variety [...] they don’t like to hear a fast-paced song comes on, then it’s just a jump for them. (S3N5, Site Champion)

Some corporate leaders’ explained differences between the two trials. They described reduced staff time involvement during the second ePCT, with more flexible protocol guidelines and preloaded playlists. Despite these improvements, staff faced issues including mailing the players to the research team for updates and extracting metadata.

We had to return our iPods in order for them to, I guess, reload music and different stuff. Cause we did have a few residents who were new and so we haven’t got those iPods back yet. But before then, it was going really, really well. (S3N6, Site Champion)

Corporate leaders also expressed concerns about sustainability as staff had not yet learned how to create the playlists; nursing homes received playlist creation training at the conclusion of Trial II.

I am a little worried about sustainability because they never did that. And so now it’s kind of throwing in a new factor here at the end… it’s not hard, it’s just yet another thing. (S4, Corporate Leader)

Quality of Delivery: How the Intervention’s Delivery Adhered to the Intended Protocol

In themes mapped to the quality of delivery domain, site champions described how they adhered to or deviated from the protocol, with reasons ranging from the champion’s appropriateness to provide the music and workflow to MP3 player logistics.

Corporate leaders and site champions described one person, the site champion, responsible for the intervention’s delivery or having frontline staff provide the music to residents. Site champions in the activities department described challenges including being the only person implementing the initiative, while nursing staff described reduced capacity for nonclinical activities during clinical leadership turnover. Site champions solely responsible for implementing the program described needing to integrate device distribution and usage tracking into their daily workflow.

I like it, but [as someone in activities], it’s just a busy job and you’re one person. But as long as the residents want to listen to it, I take it to them… They enjoy it and want to listen to it. (S3N10, Site Champion)

Involving other staff in implementing the program reportedly helped them use the intervention more often and follow the protocol more closely. Both site champions and corporate leaders reported that activities and nursing departments were primarily responsible for implementation; some reported the nursing and activities departments jointly integrating the intervention into daily workflows while others described incorporating it into other departments. One site champion described providing music to calm residents during physical therapy.

[Therapy will] say, “Oh good, they could do this while they’re riding the treadmill,” or whatever they’re doing, the bike. And they’ll bring [the MP3 player] back to me and say that [the music] seemed to help them calm down while they’re doing it… They’ll put it on them so they can relax and just finish their therapy. (S3N1, Site Champion)

Site champions talked about staff involvement, understanding, and buy-in being instrumental to ensure the intervention’s delivery. They described using various strategies to ensure the program was implemented on an ongoing basis, from educating staff to creating calendar reminders and including documentation in the electronic medical record to indicate whether music was used to deescalate a behavioral event. Staff education included information about the MP3 players, where they were located, and when to use them with participating residents, for example as an engaging activity or to address behavioral responses.

One champion discussed that assigning each resident their personal “champion” built staff-resident relationships and eased program implementation. Other site champions reported adapting the protocol to address staffing burdens by playing music on a speaker. Some site champions described putting the headphones around residents’ necks if the residents disliked wearing earphones although they acknowledged that such changes could reduce personalization of the music and effectiveness for participating residents.

A lot of them like [the personalized music]. They’ll talk about a concert they went to or back in the day when they used to have cookouts… They’ll have a different story to tell depending on what they’re listening to. (S1N5, Site Champion)

Some site champions experienced logistical barriers that prevented them from fully implementing the intervention, such as losing some of the MP3 players. A few worried about the devices getting lost or stolen, so they reportedly kept them locked. They described how this shaped the intervention’s delivery by limiting access.

It’s mostly me [doing the music], because I don’t want to leave it in there and it gets misplaced or anything like that. (S3N5, Site Champion)

A few site champions reported that they had not begun using the music due to turnover in their role. Some discussed halted implementation if the position was vacant, while others faced technological barriers such as one site champion who could not figure out how to charge the MP3 players.

Context: Organizational Culture and Events Affecting Implementation

In themes mapped to context, corporate leaders discussed the intervention’s alignment with corporate goals and both corporate leaders and site champions described implementation in the context of short staffing and the COVID-19 pandemic.

Corporate leaders described the intervention’s alignment with corporate-level goals to reduce medication use and improve dementia care. However, they discussed facility-level challenges implementing and sustaining the program in the context of the pandemic.

I loved it. I’m disappointed in the results, because I really think it has a bigger impact than the results gave it credit for. But I think part of that is because we didn’t do a great job with implementation and I think some of that has to do with COVID so if we could just get rid of COVID that would help. (S1, Corporate Leader)

Corporate leaders and site champions described residents becoming infected with and dying from COVID-19, as well as the challenges that staff faced tracking devices when facilities converted units to isolation units for exposed or infected residents.

Actually, we had a lot of, I mean we had a lot of the people in the [isolation unit], so we didn’t really, we didn’t want them to get lost in the [isolation unit] with the moves. We already had furniture missing, clothes missing from the first one and it was just about keeping track of the items. (S4N1, Site Champion)

In addition, both corporate leaders and site champions talked about staffing vacancies and shortages as key implementation and adherence barriers. Corporate leaders commented that having more consistent staffing in key positions might have improved the impact of the intervention on residents’ outcomes. They described how sites were selected and randomized to the intervention months earlier, but noted their capabilities changed by the time implementation began.

Because I think that if they would’ve felt more like it was a choice rather than a, “We have to do it right now,” I question if that would’ve made a difference because I definitely think they all wanted it but we had some facilities where, like I said, there was no administrator and there was no DON and there was no Life Enrichment Coordinator, but it was just how they were drawn in that selection phase. I don’t know. I’m curious to see if that would’ve made a difference. (S4, Corporate Leader)

At the same time, corporate leaders and site champions described the intervention as helpful during staffing crises: preventing or lessening resident agitation and other dementia-related behaviors helped save staff time and kept residents engaged during critical staffing shortages.

Also, the staff benefit from it because they don’t have to deal with the behaviors that they have. They can come get the iPods, the residents can sit down, meaning that they give them free time to go and maybe do something else instead of dealing with behaviors. So those are the two biggest things that we benefit from. (S3N6, Site Champion)

Discussion and Implications

Using the CFIF framework to contextualize potential moderators of adherence in a nursing home ePCT, we found that, while the personalized music intervention aligned with interviewees’ goals (recruitment domain) and benefited residents (participant responsiveness), they described numerous implementation challenges (strategies) and protocol deviations (quality of delivery), many associated with workforce shortages (context). Notable challenges included deciding who should administer the program and how to sustain it (intervention complexity). These findings highlight barriers and facilitators to nursing home ePCT protocol adherence from the perspective of the staff responsible for implementation. They also raise questions about the extent of pragmatism feasible in a setting where the embedded delivery of nonessential activities, such as research, is affected by staffing issues worsened by the ongoing SARS-CoV-2 pandemic.

Short staffing was an important context barrier despite our efforts to minimize burden, including research team members assuming responsibility for creating music playlists, a task that had been completed by nursing home staff in the first trial. Workforce shortages are a long-term problem in this setting. In March 2021, shortly before this ePCT began, the Kaiser Family Foundation reported that 19% of nursing homes were reporting at least one staffing shortage; by March 2022, about two-thirds through the trial, that percentage was 28% (Ochieng et al., 2022). As of 2023, staffing had still not returned to prepandemic levels and prior research has demonstrated that nursing staff ratios are associated with adherence in other nursing home-based ePCTs. The lack of nursing engagement in Trial 2 was also found in our quantitative data, which showed 40.4% of residents in Trial 1 had any nursing use of the music in the past week, compared to only 15% of residents in Trial 2. The lingering effects of the COVID pandemic on staffing likely affected nursing engagement.

Interviewees emphasized the impact of site champion turnover and suggested using a facility-wide approach to distribute responsibilities across multiple staff. Despite music players being accessible to frontline staff, site champions described feeling personally responsible for implementation, especially when other staff became disengaged. They also discussed feeling more supported when other staff were involved. A team-based approach is consistent with quality improvement principles widely used by nursing homes to pilot and scale new programs and initiatives. This approach could improve adherence and sustainability by creating processes to educate existing staff and new hires and promote the program’s ongoing use.

While the research team provided financial incentives to each corporation, their allocation at the nursing home level is unclear due to differing internal corporate redistribution processes. Further research on nursing home ePCTs, particularly those requiring significant staff involvement, should examine how financial incentives and employee recognition may shape intervention implementation. Our trials suggest that relying solely on existing resources for complex behavioral interventions in nursing homes may worsen inequities in access to innovative care (McCreedy et al., 2024). Despite removing the responsibility for staff to create the music playlists, interviewees discussed the staff burden associated with implementing the program at length. A quantitative examination of implementation fidelity highlighted differences between the prepandemic trial (Trial 1) and the subsequent trial (Trial 2). Trial 2 exposed fewer residents to music per nursing home (avg. 7.5 vs. 12.7), and residents received less music per exposed day (2.5 vs. 27.1 min; McCreedy et al., 2024). Exposure remained low despite protocol changes between the two trials, underscoring the challenges involved in balancing pragmatism with feasibility in trials involving embedded delivery of interventions by busy staff with other responsibilities Researchers examining the use of personalized music should examine the degree of personalization necessary to optimize outcomes. Reminiscence therapies aim to help manage behaviors occurring secondary to boredom or isolation by eliciting positive musical memories, with early preferred music (16 through 26 years of age), associated with greater engagement and recall than “relaxing” music (Jacobsen et al., 2015). However, the degree of personalization needed to achieve positive behavioral responses remains unclear. In this study, playlists were partially personalized, based on either staff-reports or a simple algorithm using resident characteristics. Although this process was more efficient than the fully personalized, trial-and-error process staff used in the first trial, in participant responsiveness themes, interviewees described how not all residents responded to “their” music. Some residents disliked the feeling of headphones or did not enjoy listening to music, and some staff secured devices in a central location, affecting access and use. One corporate leader—who supported both trials and could therefore contrast them—brought up shifting the responsibility for playlist creation from staff to researchers which they said negatively affected staff-resident engagement through the intervention. More evidence is needed to better understand how much personalization is necessary to elicit positive behavioral responses while sufficiently engaging staff for effective implementation.

Strengths of our study include the rigorous analytic process and inclusion of interviewees from 100% of the four corporations and 27 participating nursing homes. These findings, although specific to this trial, could inform other nursing home ePCTs using similar implementation strategies, like train-the-trainer models and reminiscence or sensory interventions for dementia-related agitation. However, this study does have a few limitations. First, we captured only corporate leaders’ and site champions’ perspectives who lead implementation, and their perspectives may differ from other staff. Second, site champion interviews were intentionally brief (less than 10 min, on average) to accommodate their busy schedules. Third, nursing home study participation was staggered, and the site champion interviews spanned nine months in 2022. The experiences and perspectives noted may have varied depending on the point in time at which they were interviewed.

Our findings underscore the importance of engaging nursing home staff as partners for the successful implementation of complex behavioral interventions. Although interviewees described the pandemic and its impacts on workforce, the pandemic was an underlying contextual theme, not a dominant topic in our interviews. Most themes instead reference logistical challenges inherent in multicomponent, personalized behavioral interventions. While exacerbated by the contemporaneous pandemic, these challenges predate it and are likely to continue. Researchers should balance pragmatism and feasibility, collaborating with nursing home staff and leaders to incorporate their knowledge of moderating factors influencing adherence into protocol guidance.

Clinical Trials Registration Number: NCT03821844

Funding

This work is supported by the National Institute on Aging, United States (R33AG057451). The sponsor did not have a role in the design, methods, subject recruitment, data collection, analysis, or preparation of this article.

Conflict of Interest

None.

Data Availability

Interview materials are included in the appendix. The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. The METRIcAL trial was registered on ClinicalTrials.gov (ClinicalTrials.gov Identifier: NCT03821844).

Author Contributions

E. M., A. R., and R. B. were instrumental in developing and administering the qualitative interview guide. J. W., G. W., E. M., and R. B. analyzed and interpreted qualitative interviews with corporate and site champions. J. W. drafted the manuscript, directed by E. M. and supported by R. B., A. R., and G. W. All authors reviewed and approved the final manuscript.

Ethics Approval and Intent to Participate

The Brown University Institutional Review Board approved the conduct of this trial (protocol number 1705001793) with a waiver of informed consent.

Consent for Publication

Each of the authors involved in this study contributed to the development of this manuscript and approved the final version before publication.