Jump to Transcript

While memory loss is generally thought of as the hallmark of dementia, behavioral and psychological symptoms of dementia like agitation, aggression, anxiety, and hallucinations are nearly universal, affecting almost all patients with advanced dementia. These behavioral disturbances are often the trigger for nursing home placement, and they can be highly distressing for both patients and their care partners. In today’s episode, Matt and Lauren speak with Dr. Ellen McCreedy, a researcher from the Brown School of Public Health who has conducted a study of personalized music intervention called Music & Memory for people living with dementia in nursing homes. Dr. McCreedy is a gerontologist and health services researcher who focuses on evaluation of non-pharmacologic interventions for managing behavioral disturbances of people living with dementia.   

More resources

Ellen McCreedy, PhD, MPH Faculty Profile 

Articles from Episode:  

Sisti A, Gutman R, Mor V, Dionne L, Rudolph JL, Baier RR, McCreedy EM. Using Structured Observations to Evaluate the Effects of a Personalized Music Intervention on Agitated Behaviors and Mood in Nursing Home Residents With Dementia: Results From an Embedded, Pragmatic Randomized Controlled Trial. Am J Geriatr Psychiatry. 2024 Mar;32(3):300-311. doi: 10.1016/j.jagp.2023.10.016. Epub 2023 Nov 2. PMID: 37973488; PMCID: PMC10922136. 

McCreedy EM, Gutman R, Baier R, Rudolph JL, Thomas KS, Dvorchak F, Uth R, Ogarek J, Mor V. Measuring the effects of a personalized music intervention on agitated behaviors among nursing home residents with dementia: design features for cluster-randomized adaptive trial. Trials. 2021 Oct 7;22(1):681. doi: 10.1186/s13063-021-05620-y. PMID: 34620193; PMCID: PMC8496617. 

Transcript

Lauren Gerlach:

While memory loss is generally thought of as the hallmark of dementia, neuropsychiatric, or what are also known as behavioral and psychological symptoms of dementia, things like agitation, aggression, anxiety, and hallucinations are nearly universal, affecting almost all patients with advanced dementia. These behavioral disturbances are often the trigger for nursing home placement, and they can be highly distressing for both patients and their care partners. These symptoms are also very challenging for direct care staff at nursing homes to manage and are associated with poor outcomes, such as increased emergency department visits, hospitalization, and increased caregiver strain.

Numerous professional societies recommend non-pharmacologic treatment strategies as the most effective means to manage these behavioral disturbances. These strategies can include behavioral interventions targeted at the person living with dementia, modifications to the environment, and caregiver-focused interventions. Despite being effective, implementing these non-pharmacologic treatment strategies can be challenging due to lack of provider training, time required to implement interventions, and lack of reimbursement, especially in the nursing home setting. Because of this, such behaviors are commonly treated with off-label use of medications such as antipsychotics. However, these medications are not preferred due to only modest efficacy in reducing behaviors and significant side effects, including increased risk of falls, stroke, and mortality as highlighted in the FDA boxed warnings.

Matt Davis:

I'm Matt Davis.

Lauren Gerlach:

I'm Lauren Gerlach.

Matt Davis:

And you're listening to Minding Memory.

Lauren Gerlach:

Today we're going to speak with Dr. Ellen McCreedy, a researcher who's conducted a study of a personalized music intervention called Music and Memory for people living with dementia in nursing homes. Dr. McCreedy is an associate professor of Health Services, Policy, Practice at the Brown University School of Public Health. She's a gerontologist by training and health services researcher with a focus on the evaluation of non-pharmacologic interventions for managing behavioral disturbances in people living with dementia Dr. McCreedy, thank you so much for joining us today.

Ellen McCreedy:

Thank you so much for having me.

Lauren Gerlach:

Dr. McCreedy was the senior author of a study titled Using Structured Observations to Evaluate the Effects of a Personalized Music Intervention on Agitated Behaviors and Mood in Nursing Home Residents with Dementia that was published in the American Journal of Geriatric Psychiatry earlier this year. We'll include the citation to this study and other works by Dr. McCreedy linked to this episode. Make sure to check them out. All right, let's dive right in. Ellen, could you start by giving us a brief overview of what your study aimed to investigate?

Ellen McCreedy:

Yes. The metrical trial, the goal of it was to test a personalized music intervention for reducing agitated behaviors in nursing home residents with dementia.

Matt Davis:

I was wondering, could you explain a little bit why you chose music as the intervention? And we're interested too if there's any theory behind how music may impact mental health.

Ellen McCreedy:

Yes. Personalized music is thought to reduce behaviors in dementia because especially verbally agitated behaviors are most often caused either... Hypothesized to be caused, very difficult to measure these things, but hypothesized to be caused by boredom or social isolation, which are common in the later stages of dementia. Pain is also thought to be... There's multiple causes of these types of behaviors, but the two main ones for the purpose of this study and that are common are boredom and social isolation.

And music, especially early preferred music, which was the target of this particular intervention, is thought to be retained in a part of the brain that's stored even into the later stages of dementia. This is something that we think about with dementia a lot. People can tell you stories from when their kids got married but they can't tell you what they had for breakfast. Well, a similar thing seems to happen with music, and even maybe more powerfully is that people can sometimes, not always, but can remember songs from early salient memories in their life they were getting married or when they were driving around with their friends as a teenager, but they can't remember songs that they liked later in life or that their kids liked, they listened to in their 40s and 50s. That's the theory, that's the neurological underpinnings.

And there's been some really exciting work, some FMRI work and other work that shows that if you find those early preferred songs that it does light up a part of the brain that's affected later in the dementia course. The idea is, and this has been shown in healthier older adults, that if you find those early preferred songs, it can reduce loneliness. Actually, in healthy adults, if you play them songs they liked when they were between the ages of 16 and 26, they feel less lonely because they're having these autobiographical memories. And so we're thinking that the same can be true in dementia because they're actually able to remember those early musical memories and they can also recall the autobiographical memories attached to them. That's the hypothesis. That's how we think it works. The goal is to find those early preferred songs, the songs that people living in the nursing home with dementia loved when they were between the ages of 16 and 26, and to play that music at early signs of agitation or at times a day when their agitated behaviors are most likely in order to make them feel less alone, less isolated, and then reduce the behaviors related to that isolation.

Matt Davis:

Is there any literature on... Have people looked at using these songs from people's pasts with cognitive outcomes in dementia? I didn't think of that until I heard you explain it that way, that it does make me wonder if music could have anything in terms of reducing memory loss or stuff like that or trigger things and help with cognitive outcomes. Have you looked at that before?

Ellen McCreedy:

There's a literature other people have looked at. Do these early preferred songs increase verbal fluency? And I think it temporarily does, for short time does. There is some not very long to these interventions. That's what we're trying to understand is how long does it work? But the things that have been reported, not by our group, but in the literature, I think there is some... I don't know if the performance on cognitive tests, I don't know if that's been looked at specifically, but I know there's some things around verbal fluency. And anecdotally and in a lot of our qualitative work around this, you hear stories, like people say that they don't usually use verbal means to communicate, but they sing the song; they can actually sing parts of the song of this early preferred music. I think verbal fluency is one that people are looking at. I don't know how much has been done to see can they actually perform on cognitive tests better? And especially in this short half-life of the music.

I think that's another piece of it is it doesn't last forever. And that's when you get to the measurement about whether to use observational measures or longer term or clinical measures. It's like we think it works, but it probably works in the moment. And then it's unclear how long those effects last on behaviors once the music stops.

Lauren Gerlach:

What exactly does the intervention look like? And how do you go about selecting the music that's used?

Ellen McCreedy:

That's a tricky part. Because people have advanced dementia, you can't just go and ask them, "What music did you like when you were a teenager?" And these are people living in nursing homes so typically have more advanced, moderate, severe dementia. You can ask a family member, but usually that's a child if they're in the nursing home, if they have a family member. Not everyone in nursing home has a family member actively involved in their care. And you might not know what your parents like when they were teenagers because you weren't around. That's a barrier.

And what the gold standard protocol is right now is to actually go and try to get a genre that they liked from the family member or from the staff that know the resident well and then go back to the billboard charts and figure out what was popular when the resident was between the stages of 16 and 26. And then play those songs for the resident and look for what call a bright response: smiling, tapping, patting, verbalizations, just increased alertness, eye opening.

This takes about two and a half hours per resident to do start to finish. It's a major barrier to scaling the intervention. And we're looking for modifications. Actually, we did a second trial with a modification that got mixed up in COVID, so we don't know if the modification didn't work or COVID really affected how we can implement the intervention on other levels. But we were looking at ways to simplify the process using some of the available algorithms. And how good is good enough? Does the billboard charts that are in the right genre, is that close enough? Or family member or staff recommend an artist they like; do we need to have these actual one-on-one testing to find the favorite songs? And that's an open question for research.

Anecdotally, I think there is something about your favorite songs, certainly. But are the songs that were popular good enough? That needs to be studied because it is challenging to find those best love songs. If you have best love songs, put them in a list for your kids.

Lauren Gerlach:

That's so interesting. Do you get a negative response to some of the songs that you play in this trial and error period as well?

Ellen McCreedy:

Negative and neutral. And it's very difficult to interpret neutral. Negative's easier, but with the muted affect that comes with the disease and just people's states at the time that you're doing the testing, are they hungry? Are they cold? Are they tired? Who knows? One day, you might see some positive response to a song and the next day you don't, and so what does that mean? We definitely see negative, but that's good. When you see a negative, you can take the song off and that's pretty clear, but it's the neutral or non-response or non-alertness that is just often a common state. And so how much is just what's going on with the resident in their lives in that day and how much is an actual this isn't my favorite song reaction? That is really hard to identify the really positive response songs. It's the intervention in and of itself almost.

Lauren Gerlach:

One thing with non-pharmacologic interventions, we always struggle with what's the right dose that we're delivering? In thinking about this intervention, how did you guys think about dosing? And what did that look like for this study?

Ellen McCreedy:

Yeah, there's no empirical evidence for dose here either as an open question. These are the open questions that are great to have a factorial trial. We're trying to get one funded, so if anybody wants to fund it. But we don't know what degree of personalization is required to see the effects and we don't know what's the minimum effective dose. The standard dose that most of the studies have used is 30 minutes. And it isn't easy to work that into a nursing workflow.

In our study, if less was effective or even less effective but could fit into nursing workflows better. Say that you're up somebody and you turned on their music while you were helping them with their ADLs in the morning, and what's the effect of that versus... It's a timing question and it's a dose question for workflow. But there's no empirical evidence for timing, and the recommended dose is 30 minutes. We are pretty close to hitting it. I think we averaged 28 minutes. Oh, 30 minutes per resident day, per resident exposed day. We were pretty close to hitting it in the first trial, but there was a lot of variation.

And we don't know for post hoc analyses look like people who got more dose did have a larger effect of the intervention, especially on antipsychotics, which is exciting. The same first author, I should say, of this paper who's a brilliant soon to be PhD biostatistics student, Anthony Sisti, is going to be publishing those results. But there's a lot of selection in that no matter what we do to try to correct for it.

We need to learn more about dose and we need to learn more about timing because those are... The way that we've done it is very tricky to implement in the real world. Early signs of agitation are very difficult to detect by CNA-trained nurses and in busy nursing home environments. If you could do it in the morning when you're already interacting with the resident in the room for ADL care and you could give them five minutes instead of 30, these things would really help scale the intervention, but we don't know yet if they're efficacious.

Lauren Gerlach:

In your study, who actually delivered the intervention? And when did they choose to deliver it?

Ellen McCreedy:

In real life when we do this and not researchers deciding what the ideal would be but what happens in real life, this was done in a real world context and delivered by CNA and nursing staff in a nursing home and activity staff. Even though we are gearing this to be an alternative to potentially inappropriate antipsychotic medication and we're hoping to engage nursing staff to think about, okay, this is when I would see this person starting to escalate; let me try some alternatives. And before I think about using a medication, a peer-run medication, in practice, because it's music, it's really hard to get it out of activities. It's hard to get the nursing staff engaged. And part of it is they have so many other things going on, part of it's just the thinking and training around how to integrate these into clinical care.

Often it was delivered by activity staff, but the target was to have frontline nursing staff use it at early signs of agitation. In practice, it was very often scheduled to be at a certain time of day, and often it was scheduled to be used at the same time of day for everybody in the program. Everybody got it when they sat down to breakfast that was on the program, which was not the intention; this what happened. But it makes us think about, okay, if that's what happened, back to the other point, it's just like, okay, let's design a trial where we do that and see how long it lasts and if we could do it either at lunch or at breakfast, depending on when you had afternoon or... You're more likely to have afternoon or morning behaviors. Will that carry over? Will we still see an effect? But yeah, in practice, it was done often at the same time a day for everyone and not tailored to early signs of agitation just because of how the workflow is. And that's what we saw mostly.

Lauren Gerlach:

Is it typically a headset and an iPod that they get?

Ellen McCreedy:

Oh, a headset. An ear-

Lauren Gerlach:

Or how do they deliver it? Yeah.

Ellen McCreedy:

A little MP3 player and headphones. MP3 player clipped to the back of their shirt so that there's no wires. And you put the headphones on, approach the person, make sure that the headphones are kept away from their ears so they can hear the music before the pressure is applied. And then the pressure, flip the cords so they're on the back and clip the small MP3 player to the back of the shirt or to the chair. That's the practice. And there's an approach to it too because you don't want to startle people and you don't want to use it too late in the agitation because it can further agitate people. There's a lot about it that is really in the approach as well.

Matt Davis:

Did your team at all track the type of music or think about that at all? It sounds like it was personalized, which is really important, but I'm just wondering if the type of music at all mattered in terms of agitation; classical versus rock or something.

Ellen McCreedy:

Yeah, we know all this. For the first trial, we know everybody's playlist and how many times every song was played. We have a lot of data for musicologists or anyone who's interested in those data. We had one student do a thesis about what... I can't think of what it's called; some quality of music and whether it was effective.

I think the thing that's really interesting to me about the play data is the weaknesses in the billboard selection approach. To back up, we had four corporations involved in the study; two in the upper mid... or in the Midwest and one in the DC metro area and one in the south. About 30% of our sample was African American or Black and not in any way a monolith among that group with their music selections by region, by site.

The biggest thing in the music selections is the site. There's a lot of commonality within a site. Yes, certainly race and ethnicity, but just... Regionality but even smaller than regionality. In the DC Metro, the site specifically, the neighborhood, the zip code, there's very specific musical choices and how much that's influenced by the staff who are deciding what their musical preferences are or what... There's lots of bias. Those are things to think about there. But it shows the weaknesses in trying to just use a Spotify or a playlist or a billboard. It's just like we did see a lot of variation in the music that was chosen. It makes sense, but the African American population that we had in the Midwest chose very different music than the DC Metro area. But even within the DC Metro area, very different music within a site by site, neighborhood by neighborhood.

I think it speaks to whatever solution we try to streamline this to really needs to engage those frontline staff that are very... They come from. And in many sites they do come from the area; not all. That's another thing that's changing, but I think it speaks to thinking about what it means to use existing algorithms, existing resources to personalize versus trying to have the personalization still happen, but maybe the activities therapist just picks songs that are generally popular with their resident. And what does that look like as a proxy for personalization?

To your real question, we haven't done a lot with the musical characteristics of the songs. Other people have tried to look at those things. And I'm happy to share some papers. I'm not that familiar with that literature about which characteristics. This is premised on preferredness, and particularly early preferredness more than any characteristic. But we do have all the songs and whatever Apple is signed there, their characteristics to be as far as beats per minute and I don't know some other characteristics.

Matt Davis:

What was your experience like conducting the study in nursing homes?

Ellen McCreedy:

Yes. We did this study. We had recruited at the corporation level, and then the corporation rolled this out as a quality improvement initiative within their organizations. It's like a strategy that we sometimes use in pragmatic trials where we're less going into nursing homes and recruiting them one by one, but actually convincing or finding corporations which are already interested in these kind of programs to making them more available for their residents and then they roll it out. We had a waiver of individual informed consent. We didn't actually consent each person in this study, we got the corporations on board, and then we made a list based on the minimum set data, which is a standardized data set for nursing homes. We identified which of their facilities, their nursing homes met the minimum eligibility criteria for the study. And then the nursing home corporation leadership either chose which of those nursing homes were going to participate or they allowed them to opt in, which was our preferred method.

Across the four corporations, there were about, I don't know, 140 or so that met the criteria and 54 ended up either opting in or being opted in to the study either by their administrator or by their corporate leadership. And then we asked the nursing homes, the staff there to... We had a launch and asked them to identify people who'd be good candidates for the program, but we never had to individually consent the residents to participate. That helped us actually be able to look at a representative group of nursing home residents because we didn't have to do the individual informed consent.

And as far as are the nursing homes representative? This was a pretty good study as far as usually if we ask for volunteers, we're going to get the high quality sites that because of structural racism, disproportionately serve more white residents. And this was a good cross-section both in quality and some of the racial characteristics. As far as ethnicity, we didn't have a lot of Hispanic or Latinx patients in this population, but it was pretty representative from that perspective in large part probably because we didn't need individual informed percent.

Lauren Gerlach:

I want to ask a little bit about how you observe and evaluate changes in resident behavior. And so in the study we're talking about today, you used structured observations to detect changes in resident behaviors rather than relying on staff reported data. Can you tell us a little bit about the assessment tool you used and why that was important to this study?

Ellen McCreedy:

Yes. We use the Agitated Behavior Mapping Instrument, the ABMI, which is part of a suite of tools developed by Jiska Cohen-Mansfield, which are considered to be the gold standard for measuring agitation in nursing home residents with dementia. It's tricky measuring agitation in nursing home residents with dementia .and there's lots of trade-offs involved in it. Trying to think about which outcome to choose and whose outcome it is and why it matters. There's several measures to choose from. And actually, we looked at three. This paper looks at one, but we actually looked at three.

One we looked at was the staff-reported agitation in the minimum data set, which is a routinely collected measure for all nursing home residents. That's what's great about it; it's already collected. It's routinely collected, but it's known to under-detect behaviors. And it's a weak look back with crude categories of how frequently the behaviors occurred. It says, "How many times in last week did occur? Every day, once a day, several times a day."

And they ask about a lot of routine behaviors which staff tend to normalize. If I'm Mrs. Jones and I have dementia and I rock or pat or tap all day every day, you're used to that, seeing me. And then when you have to answer the question, it's a little bit hard for you to decide how often it happens, the frequency. The staff normalizing the behavior is one issue, the look back period and the lack of granularity. But it is readily available and it's available for everyone, so we looked at that one.

And then there's the tool that the items in the minimum data set are based on, the Cohen-Mansfield Agitation Inventory. That's a staff interview. Takes about 15 to 20 minutes to administer for each person, but it goes through each question. Has a little bit more granular options. But it's a two-week look back. It asks how often the behavior happened in the past two week and it still has the Likert-like chunks of how frequently the behavior occurred, some of the same biases. In trying to get the same person to respond to the second one because they interact with the person differently, and that's a challenge in nursing homes. But you could argue that those two measures, they're both staff-reported measures over either the last past week or the past two weeks, might be the more important measures than the one that we're talking about today, which is the observation measure to how disturbing the behaviors are to staff, how much of a problem they are, how severe they are because they would have to rise to a certain level to maybe be reported on the minimum data set or to be captured in the staff reporting.

I think there's an argument to be made for using those measures. In effect, we chose one of those measures as the primary outcome of the study. We chose the CMBI and we did not find an effect of the intervention on the behaviors as measured by that tool. That's important to know. That has been published, and I'll make sure that's linked to because that's in a separate paper. But that is the primary outcome and we didn't find any effect of the intervention on staff-reported look back measure over the past two weeks.

Where we did find in effect of the intervention was on behaviors as observed by trained observers at standardized times a day; a time study method. And that's what the ABMI does. And so we trained observers that had good iterator of reliability that are well-trained. They went in the field with a standardized tool that has been validated. And they had set times they observed each resident. And then that's what we used when we just found that there was a change in verbally agitated behaviors, that the intervention had reduced verbally agitated behaviors compared to usual care.

Now, there's some strengths of this data and some of this measurement technique and some weaknesses. The strengths are that it gets around some of that bias that the staff have of normalizing the behaviors because the trained observers are just ticking off behaviors as they see them. And it's done at standardized times a day. And it's close potentially in time to the intervention, so you don't have to move the dial so much. You don't have to go from once a day to several times a day; it's more granular look.

But there's other biases in that they're not blinded, they're on site at the nursing home, they're not reviewing tapes with no headphones used. There's issues with the standardized observation, the way we implemented it here. We've tried to address some of those biases. But our results are pretty consistent with other observational studies done by where researchers do the observations and have found that the music works. I think it works for a short period of time. The music probably has a short half life. And that's probably why we picked it up in the observational methods but we weren't able to see it in the staff look back.

I think where it becomes clinically important is if it works and it reduces antipsychotic use... Back to your opening, even if it works for a short period of time and it can't be picked up by the routinely collected look back measures, if it does reduce PRN medication use, then we have a story. The other story there is... Or the other important thing there is even if it doesn't change medication use is the harder story and it doesn't change the frequency as staff reported over less granular look back period, like it doesn't make the difference in going from a once a day to several times a day, or from several times a day to once a day I guess so would be the direction, but it does improve quality of life for that moment... That's the other question about this. It's like what is the value of that If you're giving someone momentary relief from their symptoms? And I think that's a harder sell sometimes. It'd be nice if it reduced PRN, because I think that's the easier sell. We have some evidence that is looking positive in that direction.

But I think we shouldn't throw out things that maybe don't change the frequency over the course of couple weeks, but they do give people pleasure and temporary relief of symptoms that might improve their quality of life and, back to the beginning, reduce the boredom and the social isolation, which are the drivers of this disease. And from a human perspective. I think that matters if we're able to impact that, I think it's important, and maybe we just have to think about what we value and put as primary outcomes in the study. Maybe it changes. Yeah, I'm not sure, but those are the the thinking that I do around why we see it in the observation and not in the staff-reported measures.

Lauren Gerlach:

I'll ask maybe a little bit more about one of the specific findings. Your study found that personalized music reduced verbally agitated behaviors but had no significant effect on physically agitated behaviors. I'm just wondering if you have any thoughts about why that is.

Ellen McCreedy:

That's what other studies have found as well. And I think it's consistent with how we think that the causes of those two types of behaviors. Verbally agitated behaviors are thought to result from pain, as I mentioned, but also primarily from social isolation and loneliness. They increase in frequency when people are alone or when staffing levels are lower. And that's why that hypothesis has been formed. Where physically agitated behaviors, we tend to call them now reactive aggressive behaviors because they most often occur when a nursing staff is going in to do ADL cares or toileting, bathing, dressing, it's can... A startle effect. It's something where someone is touching or engaging in care activities with the resident and then they are physically reacting to that. Music may be helpful for those types of behaviors. Bathing Without a Battle was a study that was done that was trying to look at those types of behaviors and use music to help with those.

But I think that the majority of evidence now is really showing that these are most effective on behaviors that are not reactive aggressive behaviors and really not aggressive behaviors on agitated behaviors. I don't think we know about physically non-aggressive behaviors as much about how the music's working with those, but definitely the verbally agitated behaviors, not just our study but there's a couple others that have shown that that group of behaviors is more affected by this type of reminiscence-based intervention, which is fitting with the hypothesis of what causes them and how the intervention might work. But we did not just enroll people with verbally agitated behavior, so next study we'll enroll just those folks.

Matt Davis:

Your study found that the music intervention increased pleasure in nursing home residents living with dementia. I'm curious, how do you measure pleasure?

Ellen McCreedy:

There is a validated tool in which we used called the Observed Emotion Rating Scale, or the OERS as we commonly call it. And it looks at five different emotion types. And it gives you a face, a picture, a small picture of what someone experiencing pleasure looks like and also some examples of what pleasure might look like.

Now, some of the issues that we talked about with trying to identify someone's favorite, favorite songs are some of the same issues we have in trying to identify pleasure in advanced dementia. Classic signs of pleasure might be smiling, brightened affect in eyes. If there is music, tapping or movements in response to music, engaging with the surroundings. But somebody may be experiencing pleasure and we don't see that affect because of the disease mutes the affect or other reasons. You can be experiencing pleasure without showing bright smiles.

This is one of the measurement issues with all the emotion states around. The only way we are doing this is through observation, but what we count on in pragmatic trials is that whatever that, one, in all cases it was the same. We were lucky, we didn't lose any data collectors, so it's the same data collector doing the pre/post measurement in all cases. It's their own internal standard to some degree. And we have baseline data for each person, so that's helpful because they're within person.

And then whatever bias we think we're missing pleasure that we can't detect because there's muted affect, we would think that would be... which shouldn't be biased by what arm they're randomized to. We're probably missing pleasure in both arms because both the observers have their own idiosyncratic ways of observing pleasure, but also because we might not be able to see pleasure in people with advanced dementia. It might be muted, the classic things like bright smiles.

It's imperfect. It's also subject to bias by individuals and unblinded individuals. But we hope and we think that the bias in most cases is not differential by arm, which is a big piece of doing these kind of trials. We look for effective responses through observation. And I do think it is valuable.

Lauren Gerlach:

What are the challenges you noted was implementing this in resource-constrained nursing homes? You've spoken to this a little bit, but how do you see the feasibility of such interventions thinking about broader dementia care?

Ellen McCreedy:

Outside nursing home or still within nursing home and scaling?

Lauren Gerlach:

Maybe thinking about both. Scaling, and then maybe outside of the nursing home population.

Ellen McCreedy:

It's not really feasible to scale this intervention the gold standard way that we're describing with the songs. But it's not very scalable. I should say a little bit more nuanced than that. I don't think it's that scalable in the way that it's designed and supposed to be implemented. Both dose timing, personalization, those are the big issues. That might be the secret sauce to making it work, but also just are not scalable in a typical US nursing home post-COVID or pre-COVID, but definitely post-COVID. I know we're not post-COVID, but post the vaccine and everything.

It's just really a really necessary space to get to have this. It's a equity issue, it's a social justice issue. But it's a tough nut to crack. And I think we need creative solutions that are less good. Yes, we need to think about what are the core components of the intervention? But we just have to think what the social justice lens of bringing something and not just giving up on the setting and saying, "Oh, you have to be able to do X, Y, Z to do personalized music." But okay, let's say we gave folks an Echo and loaded up songs that the activities director told us for good songs and that a lot of people on the site liked and we told them, "Let's get everybody up on the right side of the bed in the morning. When you're in there, say, 'Play Mr. Jones' playlist,' in the morning and see what happens."

We just need to be definitely measuring the mechanisms and making sure we're looking at what's really happening, but we have to be creative with solutions that might not be... I think a lot of the zealots for these methods are really into getting those favorite love songs. And I appreciate that because I think it's really, really powerful, or that this makes such a difference because it's so different than the ambient music that it brings it... This being the headphones. It brings it into a real personal space that the ambient music that they hear often in the nursing homes doesn't and that those headphones are a really critical piece. But they're really hard to manage in the nursing home, so it's like what can we do here to make this scalable? I'm really dedicated to finding something good, but also good enough, something that does meet the essential components but it does it in a way that is really maybe less good than the fully personalized approach but it can get it to every nursing home. And not everybody likes music, so it doesn't have to be... Lots of reminiscent sensory therapies out there that are great and are growing in evidence. This is part of a suite of tools a nursing home could have. But for people who love music or liked music when they were younger, I think it's a great thing.

It has a lot of promise in the home-based setting. I think it's really, really nice to do there. We've done assisted living already. I think it is easier maybe to think, okay, this is... And there's great technology coming out too. We're testing technology that does a lot of this that tries to personalize. Depends. There's facial recognition; that's trying to read those reactions. Are those micro-expressions showing pleasure? Those things exist. We're testing them. They're not perfect, it's hard to validate them, but they're out there, they're interesting, they're innovative. If you could have something use your heart rate variability to detect early signs of agitation like a wearable. Then it also plays the music right out of the device, and so you don't need the staff there. Or you have facial recognition that starts to see they're looking a little agitated, and then it starts the music. And it can say, "Oh, they look like they like that one. Let's play more like that." They could actually do the selection based on your face.

These are products that are out there. We've seen them. We're piloting a couple. We're not giving up. I'm not giving up on the nursing home at all. I think they show great things. It's not very scalable in its current iteration, but it will be. There's solutions.

Matt Davis:

In terms of this research, I think you mostly answered this, and if you feel like you've mostly touched on already, we're just curious what's next for your team?

Ellen McCreedy:

Right now, we are testing facial recognition and software to see if we can both detect agitation, or at least signs of agitation using the software, and if we can personalize playlists based on this. And we're using another pilot to see if we can use a biometric wearable to personalize playlists. We're trying some technology pilots that are exciting and interesting.

And we are applying to do a larger factorial trial, which gets at some of the things that we've been talking about; trying to say, "Okay, how much personalization is enough or better than not doing it, not personalizing?" Different levels of personalization, different timings. Does it really have to be early signs of agitation? Can we take some of these pragmatic things that we've learned from the field and just actually see if they're efficacious? Trying to measure if the mechanisms have been activated, which isn't really done in a lot this behavioral work because it's hard because it's all the reasons we were talking about. How do you actually see if the person has elicited a memory from the song? And that's the pathway that's happening. We're going to run another trial. We're going to try it again with people who have verbally agitated behaviors that they experience on a daily basis and have received a PRN medication for those in the past few months. And we're going to try some of these less perfectly timed approaches and see if we can make a dent in a very more pragmatic approach.

Matt Davis:

Do you think there'll be teams in the future thinking more about... We're talking about audio. Visual cues from places and things that are important. Do you think the field will head that direction eventually?

Ellen McCreedy:

This there. It's already there. I'm not leading one, but on one that's doing that and also has smell in it too, which is cool. They pump smells. It's so cool. It has preferred images, preferred smells you liked. Reminiscent on all of your senses. It's there. There's products. I don't want to name any of the products because I don't know what... There's lots of innovative products out there. And we're testing. Not just me; there's other people. That one's led by someone else here. But yeah, they're already doing it; images, smells. It's pretty cool.

Matt Davis:

I'm imagining nursing home residents with virtual reality goggles on.

Ellen McCreedy:

They're doing that. They're doing that too. You're there. Yeah, it's happening. It's cool.

Matt Davis:

Very cool.

Ellen McCreedy:

Yeah. Yeah. It's like the ideas they had about just making the nursing home look old, before they were making it stop in time. You'd walk in the nursing homes and they'd be like you were going back in time. But now they can do a lot of it just virtually or in rooms. It's pretty neat. Yeah, it's wild. It's like there's so many cool things that you can only... Your imagination is the limit here.

Lauren Gerlach:

I'm always curious people's path to the things that they research and how you got here. What got you interested in this topic and in the field?

Ellen McCreedy:

I'm a music therapist originally. I came by it honestly, though. This is 20 years ago I was a music therapist. But my heart is in music. I went to music school, and then went to be a PhD in health service research, and so my life just converged again. I haven't practiced music therapy in a long time, but technically I'm a music therapist, so I come by it naturally. It's a nice thing to marry my HSR and music therapy interests.

Lauren Gerlach:

And I see that the Music and Memory website has a document that shows you how to create a personalized music playlist for a loved one at home. I have to ask, Ellen, what would be on your playlist if we're looking back to the [inaudible 00:44:48] brain.

Ellen McCreedy:

Oh my gosh, it's so embarrassing, but my three-year-old has just found my CDs. And we found a boombox on the side of the road. It was such a find. I'll just walk upstairs and hear Sarah McLachlan, and he's dancing and doing ribbons and stuff. And I'm like, "I don't know what I'm doing to my kid, but I hope I'm just ingraining in him that these are the songs that you should be putting on my playlist." It's lots of cheesy stuff like that. I'm sure it's Indigo Girls, Sarah McLachlan, all of the... Yeah, there's nothing cool about my music choices, and nothing about music school change. But there's not going to be any classical or anything, it's all going to be like girl pop from a certain time, '90s.

Matt Davis:

I miss boomboxes.

Lauren Gerlach:

Me too.

Ellen McCreedy:

Boomboxes are pretty... I almost thought, okay... Because the Echoes have all the privacy concerns, and I'm like, "Can we just get people to boombox or something?" Everything's on the table for trying to figure out the playing music in nursing homes. Yeah, it'll be something really cheesy. But I'm already conditioning my kid to make my playlist.

Lauren Gerlach:

And I'll just ask last if there's anything else we haven't covered that you want listeners to know.

Ellen McCreedy:

I think this was great. I think that people should think about what their parents love. Music is one thing, but there's whatever really brings you joy, those are the things that you're going to really want later. Some people are still sewing or doing activities that they loved throughout their life. And so just being aware of those things, asking questions. What were they into? It's amazing, they do these interviews now with a standard intake at some of these sites that we've worked with. Of course, the better sites that have the capacity to do that that ask those questions about what activities they preferred in early life. And so asking those questions now, doing those interviews like what kinds of things... Were they into music? Did they dance? Just being curious about the older people in your lives early life. That's it.

Lauren Gerlach:

All right. Well, Ellen, thanks so much for joining us today. And thanks to all of you who listened in.

Ellen McCreedy:

Thank you.

Matt Davis:

If you enjoyed our discussion today, please consider subscribing to our podcast. Other episodes can be found on Apple Podcasts, Spotify, and SoundCloud, as well as directly from us at capra.med.umich.edu where a full transcript of this episode is also available. On our website, you'll also find links to other resources we've created specifically for dementia research.

Music and engineering for this podcast was provided by Dan Langa. More information is available at www.danlanga.com. Minding Memory is part of the Michigan Medicine Podcast Network. Find more shows at michiganmedicine.org/podcasts. Support for this podcast comes from the National Institute on Aging at the National Institutes of Health as well as the Institute for Healthcare Policy and Innovation at the University of Michigan. The views expressed in this podcast do not necessarily represent the views of the NIH or the University of Michigan. Thanks for joining us, and we'll be back soon.