Innovative Use of Technology in Primary Care Delivery
Hello everyone and thank you for joining the webinar "Innovative Use of Technology in Primary Care Delivery". Before we begin, we'd like to go over a few housekeeping details before we begin today's content. First, please note that today's webinar will be recorded.
To ensure audio quality for all attendees, we've muted everyone's lines. This will enable you to hear us, but we will not be able to hear you during the webinar. We encourage you to ask questions through the chat feature, located on the lower right handside of your screen. We will hold a Q and A discussion after all the presentations have finished.
Finally, automated closed captioning is available to all attendees.To enable or disable the captions please click the CC button located on the lower left hand side of your Webex window. I'll now pass it to Matt Simpson to introduce today's content. All right, thank you very much, and welcome everybody to our webinar here today. This is the third in our series, and it's going to be on the innovative use of technology in primary care delivery. On the next slide, just want to give a quick disclaimer that the views in this webinar do not represent official views of the U. S. Department of Health and Human
Services, or the Agency for Healthcare Research and Quality. And rather highlight thaSthey are some of the research findings and methodologies used by grantees. On the next slide, I'll just take a quick moment to introduce myself. My name isMatthew Simpson, and I'm a family physician within the Division of Practice Improvement in the Center for Evidence and Practice Improvement at the Agency for Healthcare Research and Quality, and we often repeat it for ourselves as AHRQ for short. I work very closely alongside one of my colleagues, Aimee Eden who's the current acting director for the National Center for Excellence in Primary Care Research. On the next slide you can see text from AHRQ’s 1999 authorization that demonstrates the role that was envisioned for the Center for Primary Care Research should have within the government.
With this, we operationalize that into several working areas that you can see on the next slide. And with that, you can see that for point number four we are reallytalking about trying to build a more robust and systematic dissemination strategy, like the webinar series that we're hosting here today. As part of that effort to really highlight some of the research that AHRQ is interested in commissioning related to primary care and work that we're really interested in within NCEPCR which is how we often refer to our national center for excellence and primary care research. On the next slide, you can see that the focus for today is going to be the innovative use of technology in primary care delivery. This is the
third webinar and a series that we're hosting. If you're interested in the previous two recordings, they are available on the NCPCR website. This one will be recorded as was mentioned, and will be posted there as well as we mentioned. This is a series of webinars so please stay tuned for upcoming webinars. And you can stay up to date through various AHRQ channels.
All right, on the next slide, we can see the objectives that we have for today's webinar and we'll really get to see how some of the research that we've funded that uses technology such as text messaging, machine learning or mobile apps to really advance AHRQ’s mission to make health care safer, higher quality, more accessible, and affordable.I'm really excited to highlight some of these projects because I think that it demonstrates how technology can really be incorporated into primary care and primary care research in very meaningful ways. To help guide us through the webinar, I've got one of my colleagues who will be moderating.On the next slideyou can see him and he’s wearing a wonderful
suit. This is Dr. Lomotan, a senior advisor for clinical Informatics in the same center as I'm in the Center for Evidence and Practice Improvement at AHRQ. He is going to facilitate the Question and Answer session that's going to come at the end of all the presentations. Thanks Ed for being the moderator today and I'm going to pass it over to, you. Thanks Dr. Simpson. So we are delighted to have four primary care researchersheretoday. They
will each discuss their research on the innovative use of technology in primary care delivery. Let me first introduce our first presenter Dr. Anjana Sharma from the Department of Family and Community Medicine, University of California, San Francisco. And Dr. Sharmawill be presenting on her study
Experts PC Engaging Patients in Event Reporting forSafety in primary care. Welcome, Dr. Sharma. Hi. Good afternoon. Good morning for those on the West Coast and thank you for the kind introduction. I'm so excited to share some of our findings from the project, which is called Engaging Patients in Event Reporting where we're really seeking to use text messaging and tools to help improve primary care safety.Next slide. I can share some background. So as a practicing primary care, Doc, myself, I'm a family physician and as many of you are likely to agree primary care and outpatient care is the most utilized form of healthcare, and yet primary care safety often doesn't get the attention it might deserve.From what we do know, at least 5% of adult outpatients have experienced some sort of diagnostic error and at least 4.5 million ambulatory care visits may be related to adverse drug events or medication safety problems. And so our team is really passionate about the fact that in primary care now,
patients, families and caregivers, they're the ones really driving care and driving care safety. They're the ones navigating between different providers. They're the ones self-administering often high-risk medications. So we asked why aren't patients more involved in the processes to make primary care more safe?And that's what led us to this proposal.Next slide please. We were really interested in text messaging andwe really honed it on that because it is so popular,so utilized, and it seems like it can really be a tool for increasing access. And even if you look at the Pew research, for example, even very low-income individuals will have a smartphone.
And so we thought that there was really something there. I want to highlight my mentor Dr, Naomi Bardach who's a pediatrician and her work has looked at text messaging specifically for patient directed safety reporting. And she has looked at it in the pediatric care setting where inpatient pediatric patients and their families, their parents, their caregivers can real time report with a text, when they have some sort of safety concern, using text messaging. So we really were interested in that as a tool that had been shown to be really promising in the inpatient care setting. But what about for primary care and outpatient care? So, go to the next slide please. And so all of this work was really done very closely, and in concert with a Stakeholder Ddvisory Research Council. I really wanted to highlight that upfront. So everything
we've done in this work is really in partnership with our stakeholder advisors. The structure is about half patients and family members and caregivers, and then half the primary care team, so that includes family Docs, nursing, pharmacy, a doctor of pharmacy, medical assistance as well. And we really tried to have the majority of patients and caregivers and the minority being within the group being health care professionals to really address the power dynamics at play.The groups met throughout the pandemic, and we flipped to zoom. Then we sort of went back to hybrid and and I'm often asked: What do advisors actually do in this project? They helped inform all of our study procedures and protocol. They revised our recruitment materials, they piloted our recruitment protocols real time, and they have informedand reviewed all of our collected data. They have weighed in on our interpretation and discussion and then they're also co-authors, formal co-authors in our publications and
academic manuscripts.To give a little testament to their experience, I wanted to highlight a video that one of our patient advisors, Mr. Patrick McKEnna was able to record for this webinar today. Hello, I am Patrick Mckenna, a patient advisor at Zuckerberg, San Francisco, general hospital. And also a participant and Dr. Sharma’s
Patient safety notification project. In the past 17 years I've been in general hospitals, clinics and the emergency room many times. Way beyond the several 100 times actually.Mistakes that negatively affect patients do happen. But patients are hesitatant to report such issues.To have a means of reporting
problems with patient safety in a secure manner is extremely important. This is what Dr. Sharma and Amber have done. I would love to see this project become a daily resource in many hospitals. Every opportunity that allows patients to communicate with doctors and administrators needs to be offered to patients. Personally, I have complicated medical issues.
Difficult problems arise during the care, and it's not easy. To communicate them, this project allows me to communicate what I find difficult to say. Thank you, thank you, Patrick.I'll talk now a little bit more about what does this actually
lead to? So, I think on the next slide, I share a little more about that and so this is really. We're in process, so this is a really iterative process. I would say it really started with our stakeholders advisers, sharing journey maps about their own experiences with a primary care error or adverse event. And then also some end of focus groups and qualitative interviews that were also funded by, by AHRQ to lead to the actual intervention.Which, you know, is really around the core finding we found was that communication was key. And so it was so often
that there was some sort of concern or symptom or question mark about something and not clear, ways or means to be able to convey those concerns in a timely manner before an actual adverse event occurred. And so that was what really let us to think, okay, really based on what we're seeing so far there is really something here as far as text messaging as a tool to enhance communication. And so the thing, on the next slide, I have a little more and so then we wanted to trial the fixed tool that was from Dr. Bardach and since that tool is already built, we did a quick and dirty cycle really thinking about these user centered design principles. So really, let's try this out. Let's get this going. Let's see what it does. And so that was with adult patients at our public hospital,
here, the general hospital.These patients were English speaking adults who did own a smartphone and had some sort of personal experience with insulin, opioids or blood thinner medications. Because we have found in precedent research that these are medicines that are much more associated with adverse events or safety events in primary care. And so then, I think on the next slide, and so here's where this is, really what is the tool? What does the tool actually do. So participants both in this inpatient phase, and then in our outpatient adaptation receive a simple text message that asks, have you experienced any safety issues today? And so if, so you would either respond a yes or no why? Or an N. And if you and so you would just want to know if you didn't have an issue if you did have an issue, you can click on the link and the link leads to a mobile online reporting tool that's HIPPA enabled and would enable patients or family or caregivers to describe what kind of event had happened and also share their narrative. And we provided some categories for them, and then as well as other, and then also an opportunity to provide positive feedback. So an opportunity for them to say what went
well, and so that is really the, that's really the architecture of the tool. It's really designed to be as simple as possible and as brief as possible while still trying to get good information, and also really enabling patients to share their own story. Okay next slide. So then we trialed this tool among 10 patients, and each of them got this daily text message for a two week study period.
You can see that the cohort was fairly diverse, racially, ethnically, and by age and we found a number of reports in a number of categories. So, 16 reports among 10 patients for 2 weeks, which I can tell you right now is definitely more than you might find from our typical built in safety reporting systems for primary care. However, we found that not one, but two of those 10 patients did report, unfortunately, some sort of issue related to chest pain and that really caught us off guard because we had done extensive counseling and orientation with this tool to not use it for emergencies. The whole point Of this is really to be more of a safety monitoring or concern tool rather than a rapid response. And so we, even in our orientation with this phase 1, we specifically use chest pain as our example and to use teach back, which is highly validated as a tool to help to teach new interventions.And we said to patients if you had chest pain, what would you do? Still with that, we hadtwo incidences of chest pain, so this is part of research, right? So, we then we halted phase one and we really went back to our team. We went back to the stakeholder
advisors and so I think that's on the next slide. We wanted with our stakeholder advisors. We developed interviews, we asked each of those 10 patients who trialed the tool what they thought and sothe benefits we found that the tool is really easy to use. It was accessible, there might be some questions about where does that information go?And we also got some good insights into why it might have been used for an emergency and so I have these quotes here from our notes from our usability interviews. But really, my takeaway is that especially in the public health
folks have a lot of concerns that it can be really, really hard to get in and to get heard and so a rapid text message that you get really feels like you're being cared for, in a quicker way. And so the timing around this tool, I think is something we really still need to think about. We, as far as what to do next working with our input from these interviews, and from our advisors, we made dramatic changes to our orientation protocol. We added in multiple stop gaps along the actual portal of the mobile reporting tool to confirm this is not an emergency and to click it. And then we also got another round of display and more interface feedback. Okay. Next slide,please. And so then with this revised tool, we then had - cause we did really want to hold safety at the core. We said, okay, let's have all our stakeholder advisors trial this out themselves and see what they say. And so that was our next phase of trialing where we had our doctors on the team, and the nurses also tried out the tool. So the next slide we have those findings. So this is just a little bit of showing of what changed. So we also changed some wording and we really made it clearthis is that medical safety issues is a really clear clear issue, clear aspect that we wanted to.
Our group worked a lot on wording really And a lot on visual cues. Okay next slide please. So this is another example, where you can see, we added a lot of detail about what does emergency actually mean and we added other stop gaps as well. In this in these updates to really try to make it more safe. And then the next slide, we also changed the categories we wanted to recognize that the mental model of safety in primary care and ambulatory care is definitely different than in patient care. And so honing in on the categories you can see for the experts tool that we thought were most relevant and came again from our stakeholder advisors.
Okay, the next slide. So then we had the stakeholders trial it out and on the next slide. We had 9 folks try it out again for 2 weeks cycle again, getting a text message every day again, just like other participants and we also have them use the system usability. Which is a well validated metric to assess how usable a new innovation might be. And we were scoring in the 70s. So 70 s is
really considered in the good range. And so we, as a group did think that this was an appropriate level for us to proceed. We also wanted to check the digital literacy of our participants and so that is the health literacy questionnaire and as you can see that patients and caregivers did have somewhat lower digital health literacy than the than the providers and healthcare staff. So that's reflected there. Okay, next slide. And so then here, so we also added in based on that round of the stakeholder feedback, even one more change to this, where we added this link where it says here and that was a place that people could even see what would be some examples of emergency situations. We also fixed a little bit of some of the navigation that was glitching or a little bit confusing at the very end of the tool. So it'd be clear what to do if you needed to exit, versus if you wanted to put in multiple reports. Okay, next slide. So then we felt like we had done a lot to try to improve the tool and improve safety. And so then we're now in phase
three, with a new cohort of adult outpatients, with similar inclusion criteria, adults who have a smartphone who have experience with insulin and blood thinners or insulin or hypoglycemic medications, blood, thinners and opioid medications. And then on the next slide, we are in deep in the midst of phase three testing as we speak. We have a goal of 30, at least 30 participants. We have reached 8. We have 17 reports today, and I am very pleased to say we have no emergent uses of the tool so far but since we're actively collecting data, I'm not sharing the nitty gritty so far um. But we're really eager to see what we learn and what happens next and reviewing that back again with our Stakeholder Ddvisory Council.
And then next slide, so, as far as benefits and challenges really grateful for the work done in partnership with stakeholders, all of this work is at a public health site that serves a diverse, low income population of patients. And so this, we have high confidence that the tools that we do develop are useful for a population that is often neglected in research. And I will say that as far as the challenges, you know, this work, thoughtful work, user centered work, it does take more time. It's not as easy as someone just working alone in a room.And so it has taken us a lot much as much.
lonnger to rabbit a tool that we think is effective for our patient population. So I think I think that's enough for this this slide. And then on the next one, I really think we're thinking now about the right fit for this tool and the right setting and the rightpatient population. And so I think there's a lot to be said for maybe this tool is not for everyone, but for specific high-risk populations in a specifically vulnerable time in primary care. And so that may mean right when you start a brand new high risk medication, it might mean right when you're discharged from the hospital and that really, you know, acute period of vulnerability rightafter discharge. So, I think those are some really specific applications for this tool that could be coming next. And then, as you can imagine, we really want this tool not to be. another add on ideally, it would be integrated into existing systems, be that either the electronic medical record, or within our existing incident reporting tools that each hospital has these means for where the staff can put in a safety report. So, how can patients may be engaged to do reports in that same fashion?
And then, and then there's also this piece where, you know, seeing how patients really do want that closer contact and more real time feedback. Is there a role herefor AI, is there a role herefor whatmy mentor, Melissa Carr calls chatbots for good.You know, where we can maybe have for some of these text message iconcerns,ss there a way that we could actually enhance feedback to patients in a way that is safe, but also would it not necessarily incur more burden on our already overburdened health primary careteams. So these are all thoughts I would love to hear what folks think. And also what the copanelists might have to say. And so just want to express deep, deep thanks for the entire study team.That's my contact information. If you want to save it feel free to reach out and just big gratitude to our advisors as well as our study staff. And my mentor is Dr. Bardach Thank you so much. Thanks very much Dr Sharma.
So we're going totransition to our next speaker. Next slide.But maybe first, I’ll remind folks you can submit questions to the chat. So don't be shy. Please submit your questions to the chat and we'll address those at the end of all the presentations this afternoon. Please join me in welcoming our second presenter, Dr Adrian Aguilera from the University of California, Berkley, School of Social Welfare and the University of California San Francisco, who will present on his study, Improving Diabetes and Depression, Self management, via Adaptive Mobile Messaging. Thanks so much for having me, we can go to the next slide and get started. So, first, I'd like to acknowledge the Co-principal
investigator on this project, Dr. Courtney Lyle, who's at the Center for Healthcare Policy and Research at UC Davis. Next okay, sotoday I'm going to talk a little bit about our, the DIAMANTE project, which stands for diabetes and mental health, adaptive notifications, texting study. The impetus for this project was an interest in addressing co-morbid conditions that influence each other negatively being diabetes and depression, which are often seen in primary care. And what we know is that interventions for depression and diabetes are siloed so you often get treated for one or for the other, and we are motivated by finding core mechanisms for these interventions.
For these outcomes one of those is physical activity. So we know that improving physical activity among people with diabetes improves diabetes outcomes and increasing physical activity for people who are experiencing depression also improves depressive symptoms. We're motivated to use mobile health interventions to try to scale out what we can do but we know that a lot of work needs to be done to make them more personalized to better integrate them into care and to develop them with vulnerable and marginalized populations. Who often get lift out of the development side of the equation, next. So the primary aim as I mentioned
is to increase physical activity among patients with diabetes and depression. We aim to do this by first, collecting step counts from individual smartphones, and then using that data to send personalized messages using a reinforcement learning algorithm. So, this is a type of machine learning algorithm, which I'll talk about a little bit in more detail in a bit.And the way we plan to personalize is to target individual specific motivators. So there's some, there's different things that people need to engage in new health behaviors and also discuss that in a bit as well. Next. This is an overview of our DIAMANTE app. So this is how we, collect data
from participants’ smartphones and their step counts.There's an example of the app, which shows two individuals how much they walked over time. And then we send them text messages. First, a feedback message related to how much they walked in different ways. We can either say you walked less than your goal. We'll give them the actual number of steps that they did and that's another thing. We vary a little bit as well. And then the follow-up message is a motivating message trying to get folks to improve their step count that day or the next day going forward. Next we generated content for the motivating messages, using the behavioral change framework and so we generated messages related to capability. So these are messages that are, we might often
think of as educational type messages. So, how why to engage in physical activity why is it important How is it related tomaybe diabetes outcomes to feeling better motivation we framed as self-efficacy so getting somebody to move. So “you can do it” kinds of messages. And we also did some user centered design early on in the study for about a year, which is really important.
In getting to know what was relevant to patient's lives. And here a lot of folks wanted inspirational type messages. So we really ramp that up when we develop the content. And then the third category is opportunity and so opportunity is more about thinking, helping folks think about how to engage in the behavior. So putting sneakers next to the bed so in the morning going out to walk with a friend, things like that.
So the reinforcement learning algorithm is a type of machine learning algorithm. In this case, there are many, but there are three broad types of machine learning algorithms. One would be supervised, which is using existing data that alreadyhas been collected and is learning based on that. Unsupervised is kind of learning on the go and reinforcement learning, kind of learns on the go as well but is very specific to what we call a reward.So technically define how software agents ought to take actions to maximize the cumulative reward. In our casethe cumulative reward is step count. So the equation is trying to maximize the amount of steps
that it's predicting the patients will be taking. Next so dig down a little bit into this since this is probably not familiar to everybody. Here's some important terms when we're thinking about reinforcement, learning algorithms. So the first is action variables and this is what the algorithm is doing and so we asked for them to do a few different things. One to vary the feedback that people got. So the message is related to what they did yesterday. So we might just say you walked 5,000 steps and just leave it at that. We had a message that would say you walked more or less than your. goal which was set at the beginning of the study, or we would send a message saying if they didn't meet their goal, kind of, you can do better. Maybe a little bit more of a push.
There was that motivation message, which I mentioned, which we took from the COM B Framework and then there was a time of day since it might that might vary based on when people are more likely to read a message and really act on it. There are contextual variables and these are related to an individual's character. So, let's say their gender, their sex, their age, other factors that are related to individuals, and also their behavior so what they've done in the past week in the past month, et cetera, and what's going on in their surroundings. So, for example, we included weather data here. So if it was raining, we could guess that physical activity is likely to be a little bit lower. So the algorithms are taking this into account. The reward is the outcome as I mentioned physical activity. And the model that gets put together is how these things are predicted how we predict the potential reward might be as a function of these action and contextual variables. Next so, just to put it all together every morning, an algorithm is retrained on data of all participants.
So we had the action variables. The contextual variables are put together to come up with a reward. So here are some examples of some calculations it might make, so it might choose to send out no message a benefit message, self, advocacy, motivating message, or an opportunity message and then it chooses a random message from whatever category it's deciding to try that day.It's important to note that the actions aren't always the best ones, so it's not going to select a benefit message every single day, even though that would maybe predict to be a little bit higher because it's going to try out different combinations of messages, the effect of the similar types of messages may wane over time, so these things are taken into account.
Next all right so now getting into the actual study. So the participants on the setting were low income patients from a primary care clinic at a public hospital. Same hospital that Dr Sharma’s patients are at in San Francisco. We worked with English and Spanish speakers with diabetes diagnoses and elevated depression symptoms. There were very digital literacy levels. So people were not all, you know, super users of apps and text messaging and so on. So we have to do a little bit of training and onboarding. And part of that was related to doing the one year design phase to really take into account the digital needs of patients. We ran into a little bit of an issue with this covid 19 thing, which started right around, our recruitmentstart point as well. So we had to turn to social media to expand our participant pool
where we got about half of the sample. We also sent step and data to primary care physicians via electronic health record messages. So our integration could be better, which I will talk a little bit about at the end. But there was some integration and I think importantly, patients. All these messages as being part of part of their clinical care generally speaking, next.
Hear you can see the trial design where participants were in three different arms. The adaptive arm is where they were receiving those reinforcement, learning messages and weekly message, a weekly message, asking them about their mood rating. The random arm was folks just getting the same messages as the adaptive arm, but randomly. So there was no algorithm that was turned on messages were just sent at random.And then the control group just got the app and was asked to track their steps and then receive the weekly mood rating message.
Next so here, the numbers of folks who participated in the intervention. I think a few things to note here are, the average age was around high forties, low fifties. A little bit higher female participation, but it's actually not too bad when we look at other research studies that tend to have a higher female participation, a significant Spanish language speakers. And a good mix from a racial and ethnic standpoint. Next. Again, given our focus on diverse, underserved populations, education levels also range quite a bit with a fair number with only a high school education, or even less and then also age and employment status.
Next so, before the study folks were averaging somewhere between. 3 to 4,000 steps a day, which is fairly low. So it is a population that could stand to use some help improving step counts over time. And next, so now to the final outcome, so we did find that the adaptive arm, indeed had higher step counts over the period end of this study. So the study lasted 6 months. And their steps increase over time, whereas for the random and control arm, they were relatively flat or even slightly declining. So that's a promising result showing us that personalizing these messages can provide some help in increasing step counts for folks with diabetes and depression symptoms.
Next, so, some of the implications are broadly that text messaging in particular, it's people talk about it as a workforce in digital health. We saw in the previous study by Dr. Sharma was also using text messaging. It's not the most fanciest, but it provides a sense of connection and ease of connecting with people that relies less on using an app, which requires much more engagement. Personalized and adaptive interventions may be particularly effective at increasing activity and we know that this is where things are going. So we started this project back in 2017. I believe. So. I'd like to think we're a little bit of ahead of our time, but the move is more towards personalization now that we're moving towards large language models. For example, we're continually moving in this direction and I think it is a direction to supply properly as Dr. Sharma mentioned, for good. It can have a positive impact and our hope is that the ideal is that it reduces existing disparities.
An area to continue to work on is to better integrate these tools into care and more seamlessly into the electronic health record for improved implementation. There is a sense that people are more likely to engage in these digital tools when they feel that it's part of their care that it's coming from their doctor's office and their care team, versus if it's just a standalone program that's separate. Okay, I believe, I think that's it. Nevermind, yep, thank you very much. Thanks. Thanks. So quick reminder for folks on the line and watching, please, do think of questions and submit those to the chat, we'd be happy to take those after all the presentations are done. Let me now
introduce our third presentation for today. We have two principal investigators Dr. Ryan Coller from the University of Wisconsin, Madison, School of medicine and Public Health and Dr Nicole Werner from the Indiana University School of Public Health Bloomington.They will present on their study mproving medication safety for medically complex children With M-health across caregiving networks. Welcome Dr. Coller and Dr. Werner. Thanks so much and a huge congratulationsto the Co-presenters for their really exciting, work and we're thrilled to be able to share ours as well. I'll get us
started off on the next slide with a little bit of background. There could be a misconception that children, don't necessarily use a lot of medications in general. But the reality is according to the data that more than half of children in the United States take at least 1 medication per week. And as many folks can imagine a lot of those medications are given in formulations that can be difficult to give accurately.Such as liquid medications, or the child kind of struggling to take what's given to them.And alongside that we see that the vast majority of medication errors are happening really in the home and community environments and this statistic at the bottom is quite staggering.That every eigh minutes, a child experiences a medication error during routine care at home.And on the next slide, we're zooming in on
a population that Dr. Werner and I have been particularly interested in in working with on this topic, and that's children with medical complexity - that you'll see abbreviated as CMC. And if you haven't heard of that population before the way that they're typically conceptualized is in the figure at the top, right? Where children with medical complexity live at the intersection of four related Concepts. They've got chronic conditions; severe, functional limitations; high service needs and healthcare use.An image of a child's daily medication regimen who's got medical complexity is photographed at the bottom, right? Where you can see that there's multiple different formulations, including injections, topical Medicines, liquid medicines, and tablets. aAnd this population in general
takes medications that are particularly high risk, because there are frequently drug interactions and the space between a dose that's effectiveand a dosethat's toxiccan be kind of narrow. So this group is particularly vulnerable to medication safety challenges and you see that in some of the healthcare use datathat's at the bottom of the slide where children with medical complexity have five times higher odds of having an EDvisit due to an adverse eventthan other children, and many of those lead to hospitalizations. On the next slide, what I wanted to really illustrate here is some work by a collaborator of ours Dr. Feinstein and his team that has put data behind the complexity of medication administration that I just was referencing. So the figure on the left is
a way of looking at a medication regimen for a child with medical complexity, using an index called the medication regimen complexity index. And what that does is, it's a composite that really brings together the number of different dosage forms, frequencies, and special instructions that span the child's daily medications that they receive. And the top part of this figure shows that children, whether you look at low index scores or high index scores,rthey are ranging between 4 and 8 different forms, four and eight different frequencies, and four and eight different special instructions across their regimen. And at the bottom, the really striking point of these data are that children, even with the least complex regimens who have medical complexity are being administered 20 different doses throughout the day. And that number jumps up to 50 Different doses of medications given in the home when they've got the more complex regimens.The figure on the right shows a
different, and very understandable consequence of this complexity. So the histogram bars in white are parent self-reported mastery of different medication related tasks. So indication knowing indication the dosage and the measurement of the child's medicines and the black bars are the demonstrated mastery when observed. So you can see a huge discrepancy between perception and demonstration of medication mastery, which is again, understandable, given the complexity. On the next slide, there's an element that hasn't really had a ton of attention yet in quantitative research, but we all, I think, qualitatively understand that there's another unique attribute of children with medical complexity that is at play here. So, Lily in this
figure, which is called a care map, is commonly used in understanding the world of a child with medical complexity. Lily's at the middle, and she's surrounded by her family. And Each of the hub and spokes surrounding her areas and different colors represent different domains of her life, different aspects that really are critical to her thriving. There's a lot, and each of those little bubbles can represent a person or a service and many of those might be places where Lily needs to have medication given to her or care tasks given to her. So, you can imagine there's this really critical need for those different people and settings to be coordinated and communicating with onr another in order to do this well and give her safe care.
So, on the next slide, a really nice recent publication I think distills down a lot of where our work has led us. zThese are primary drivers through a key driver diagram to really focus on safe medication administration in the home for children with medical complexity.And the authors highlight medication management, care coordination, and patient and family engagement, which is exactly what our work is trying to achieve.
So, on the next slide, I'm going to just highlight two gaps that lead us into our technology solution. So, the first is really we lack tools that can address those challenges for the sheer complexity of medication administration for CMC caregivers.And the second is that no tools really exist, well, that are designed for CMC and their families to support coordination across the network of Folks who are involved. And on the next slide Dr. Werner will take over. Thanks Dr. Coller. So our goal was to address those gaps that Dr. Coller just highlighted through a family facing mobile health information
technology that could manage medications and manage care and coordinate and communicate that care among the caregiving network,all those caregivers that are involved in care. And we wanted this technology to be designed by the families and other stakeholders that would be using them. So, through some initial funding through the maternal child health bureau mobile health challenge, we did engage with families as the designers to create prototypes of an initial solution focused on tube feeding to start. So you can see here on the screen some of the photos of us working with those families across co- design sessions, starting kind of with sketching and creating these paper, low fidelity prototypes and moving all the way toward an actual usable prototype. Excuse me so, I’d like now for you to just hear the words of one of the parents that participated in a Co-design about how they think this digital solution they designed is going to influence the safety and quality of their child's care. “Because, like it is, it's a one stop, one area where they can have all this information and have that input it and make life simpler and hopefully cut off on human errors in between caregivers and spouses. even.” You can go to the next slide.
So, building on that foundational work, built on that foundational code design, we were able to develop the app and test that initial prototype with families. And in that feasibility test, familys reported that Tubes at home, which is the original version, was usable and useful and also feasible for use over a period of time in the home, next. And so we were able now to build on that foundation of work for the current proposal that we're talking about today, where we're seeking to adapt that initial Tubes at home prototype, using again co-design with families and other key stakeholders focused on supporting and improving medication safety for the child among the caregiving network. And we're calling this Meds@home. And after we developed the app, we're testing the effectiveness of Meds@home to improve medication administration accuracy in a randomized trial with families in the home. And go to the next slide.
So, in our first aim, where we're adapting the design of tubes at home to create Meds@home, we convened three groups of designers. So primary caregivers, which were parents of children with medical complexity, secondary caregivers, such as school nurses, home health nurses, and clinicianswho care for children with medical complexity.. So, we had nurses, complex care pediatrician, pediatric clinical pharmacists, and someone from a home health represented. And we brought them together for a series of co-design sessions so you can go to the next slide.
And here is just a kind of table schematic of the code design sessions that we did with these three different groups of stakeholders. And so we held four design sessions with the parents. And two design sessions each with the secondary caregivers and the clinicians, ending with an evaluation, individual evaluation with the family so that they had kind of the final input on the design. And we integrated the secondary caregivers and clinicians in between the code, in, between the co-design sessions of the parentsso that we could make sure we were getting different stakeholder design. But that the parents were kind of driving as the main users were driving the major design process. We used weekly meetings with our team to have consensus discussion and plan for the next session and work with our user experience designer and software developers so that we could make prototype changes in real time in the sessions integrate all the feedback or the all the designs that the co-designers created in session and have those ready for the next session. So it was a really a
creation and development happening iteratively and in real time. You can go to the next slide. In our sessions, what we focused on is aa five stage process that we use in our research for co-design where we start with identifying the problem, and we generate solutions, converge on those solutions, prototype, and evaluate. And so the three key challenges that our stakeholders and parents, secondary caregivers and clinicians identified were giving the right medications at the right time. Communicating about medications to the rest of the care network. All right, so when was the medication given? and parents often talked about this leading to double dosing and or missing a dose.And accommodating complex, medical
needs - so how to integrate all the other complex care needs, symptoms into medication management and the communication of that medication management to the rest of the care network. You can go to the next line. So what we did is we took those throughout those sessions. We took those initial challenges and the Cco-designers generated specific design requirements to address each of those challenges and converged upon and prototyped this design. So you can go aheadand play the video. This app is designed to help manage complicated care giving tasks and medications. The home screen shows important alerts, care routines for the day, and medication refill reminders. The basic functions can be found in the task menu at the top left of the screen.
Tapping the icon in the top left of the screen opens a menu with several options. The medications and routines tabs allows primary account holders to add, edit or delete medications and routines. The calendar tab shows upcoming and previously completed routines week by week. The notifications tab displays notifications about upcoming and past due routines as well as refill notifications. The my profile and child profile tabs display important account information and can be edited at any time.
Check out the other videos to learn more. Thank you, you can go to the next side. So now that the designers, the co-design teams have created this prototype, we're testing the prototype that they created with families in a randomized clinical trial. And our goal is to evaluate the effectiveness of the Meds@home app they created on medication administration accuracy with the hypothesis that medication administration accuracy is going to be improved with the use of Meds@home, both in the effect on primary caregiver medication administration accuracy and then also the effect on secondary caregiver medication administration accuracy. I'll turn it back over to theDr. Coller. Great on the next slide. So, we are, we'll be excited to report back results when they are done. We're thrilled to have the opportunity to conduct this efficacy study.
But we also recognize that as a single site study, there's limits to what we can learn, particularly around implementation. And so we're hoping for a subsequent step to be able to look at both effectiveness and implementations together. So that's one of our immediate next goals. And then beyond that, or perhaps and sort of simultaneous with that, there's a lot of things that can be easily built into this platform that we've got our eyes on. So, action planning for example, if a child has an acute health crisis, a lot of times, there's need to respond quickly and accurately with medications and other caregiving tasks. So, we're looking forward to the opportunity to build some of that functionality into the technology platform.And then as we've been hearing from the other speakers, and in other research,
the opportunity to use this information that's being collected by families and other caregivers throughout the day gives us the opportunity to use that information in order to monitor health as well as make predictions about outcomes that might be impending. So we're looking forward to the opportunity to build out more of that as well. And then as it relates to the connection to the EHR, I think that's a nice segue into our final slide. Which is about integration into primary care. Interestingly, when we began design sessions, the question about integration into the EHR came up, and it really wasn't prioritized by families as something that we should focus on right away, because they really wanted tools that they could use themselves within their own environment.So there hasn't been that focus to date. But what we think this work really allows us to do within the context of integration into primary care is shed light on what's really others have referenced as the invisible system of care, particularly for individual who have multiple complex and chronic conditions. It's a very sophisticated system of care that families and caregivers have to develop and manage and tools like this
allow us to make that work less invisible. Gives us the opportunity to study what's happening and implement supports to that. And really create more of aseamless line between healthcare settings and settings at home and in community. We think HER integration, when we do go down that road with this technology, is going to open up new opportunities for communication between these caregiving teams and clinical teams. And we think there's opportunities to promote a lot of what happens in primary care using the technology. So
for example, in pediatrics, we have bright futures, as a anticipatory guidance framework, and the opportunity to deliver recommendations, ask questions, connect to resources asynchronously from clinic visits in a more automated fashion using technology like this would be one potential waywe could see leveraging this to achieve better integration into primary care.And I already mentioned remote monitoring and at some point we'd love to apply machine learning to identify risk and intervention points. The key unknowns we've listed there are things that we'd be happy to elaborate on more in the Q and A as well as open questions that we're still wrestling with ourselves. And on the next slide, I will just say Nicole, and I wanted to thank our Co -nvestigators listed there, our research teams doing the heavy lifting and our community partner Barbara Katz has been really a lifelong advocate and policy expert and partner with us as an individual and through her work with Family Voices of Wisconsin. Thanks to our family and community co-designers and our software developer, mobile applications as well as AHRQ for supporting this work. And with that, we will say thank you and move to the Q and A. Thanks very much,Drs. Coller and Werner. So now we get to the fun part around the Q and A.Again for those in the audience, if you have a question for
our presenters, please do post some of those in the chat and I can start this off, as the moderator I guess I will take the moderators prerogative and and build on some of the things that Dr. Coller mentioned in his last slide.Andit has to do with integration into primary care systems. So, maybe we'll start with Dr. Sharma. You alluded to it also. Is there a chance to expand a bit on your thinking of how your technology might integrate into primary care systems whether it be EHRs, or other technologies and in particular, if you can comment on what do you think the workflow integration or applications might be if it were integrated into the EHR that would be great. So we can start, Dr Sharma. Thanks. Thanks for that. Good and tough question I think The difference I think withours, our challenge is that even safety incident reporting is not well integrated with most EHRs. And so I think that there's a divide in
a lot of the patient safety reporting tools that we have, even from the clinician side, that don't talk to, like, Epic or other electronic medical records. So I think that that's like an additional challenge thatwe are being faced with. The piece I feel we really need to solve would also involve integrating with pharmacy. So I think a lot of what and I’d be curious to hear what the That what Drs. Coller and Werner think as well, you know, because I think that it's actually the triad interface
between the clinician, the prescriber, the pharmacy and the caregiver that doesn't talk to each other. So I think independent of our text messaging tool. I think there's a lot that's broken already that we need to fix. But I think what would be an easy and direct.
idea for this tool I mean, the way that we have been handling messages is in our small team, I'm a clinician, I've been reviewing the events and I've been integrating it myself into the electronic medical record by through, you know, making a research encounter or a telephone encounter. And so I think if there was a way that the tool itself could just trigger , my chart message - we use epic here. I have no affinity specifically to Epic at all. But just as an example that if it could, if those could be safety reports that could then just go in as a patient message,that's one option because that would triggera nurse triage.
Gosh, I think the elephant in the room is that at some point patients will want text messaging with their HER.You know, I think that I don't think most primary care Docs want that. I don't think most clinical teams necessarily want that, but I don't see how, I don't see how that doesn't happen in the future. So I think that if that was a, if that was a mode, this would be a way, sort of have a safety flag on safety text message events that I think could be feasible. So, that's all I got. Thanks. All right, that's a great, great, great answer. And it's a good reminder. You know, not I think Dr Collermentioned this also - not all applications or technology at the outsetshould be thoughtof as.
integrated into the chart.Or some some pros and cons every Every application,every platform. So Dr. Aguilera, do you mind just expanding a little bit about what you think your platform might have in terms of potential for integration or not.
Yeah, I think the challenge with any of these interventions you know, these digital interventions being integrated into care are making sure that the information that gets sent to providers is important, actionable information. I think, you know, the worst thing we could do is just send a bunch of raw data. Nobody wants to do that. And I think first we need to show thatthese things work. And then the part of an implementation outcome to look at His ow we can then deliver this information. So, for example, we delivered it via emails, you know, messages in the records providers showing, a basic graph of step counts over time. There are more sophisticated ways to do that. But the real challenge is, you have to build these tools into theE HR, which are often hard to access, because, you know
we can't go in and just build things. So I think the key is getting things in ways that are digestible. I also think we need to do work talking to, primary care providers, asking themwhat do they want to know? What do they want to see? So I would think, you know, physical activity is something that providers would like to see, but they probably want to see it in a digestible way, so maybe just generally what has been the trend to physical activity for the past few months right? And then have a conversation about that, and I think we could apply this to some other things. You know, we already asked for things such as PHQ9 ratings and other things. So, again, we really need a lot of work in terms ofhow is this information delivered? Thanks very much. Dr. Coller and Dr. Werner any comments around,
more comments from technology integration or work flow integration. I can start I, I think thatone of the challenges that we didn't allude to. I mean, some of the practical things were deliberately trying to get a tool into the hands of people that are going to be sharing it with others in their world. And so, I think the barriers around, like, who has access to what kind of information is something that's a unique challenge that we haven't entirely addressed when it comes to if you've got an EHR that's potentially talking to the app if they're talking together.
So that's something that's been interesting. And then the other thing is many of our patients have the potential, at least, to be seeing providers at multiple institutions and so that creates a lot of additional challenges of integration of a single technology, potentially across multiple institutions. So, those are those are just two of the things that cross my mind. And then, I guess the other that's pediatric specific is around, um, privacy as you know, the caregivers are really the owners of the information and decision making but that obviously switches as the child becomes a teenager and an adult. And in our population, we have a lot of folks who also might have guardianship. And there's just a lot of layers of complexity that need to get, I think addressed. Particularly when it relates to integrating with health systems structures.So those are barriers where we foresee challenges. We foresee lots of
opportunities to work on them. Nicole chime in, on anything else I forgot. Just a quick chime in about not that you forgot anything. I thought you covered it and just kind of speaking to the other all three of the great comments that, like, we really need to think about not just what, what do the clinicians or primary care clinicians want, or need, or what's useful to them but also, like, how are we going to fit it into their system? So, not just in the EHR itself, but in, you know, where in the workflow would this information be provided? Where in the workflow would they be using it? And indicating it to the patient or family, and really thinking more, like, stepped out from the technology like what's happening in the room, what's happening in the clinic and how are we going to integrate that more from a systems engineering perspective? Yeah, those are great points and I think we'll build on this a little bit and Dr Aguileara, you touched on this already a little around, you know what do you think clinicians might think about technologies like this, if we are able to successfully get them into the EHR and if they're appropriate for that. And as you think about those workflow questions that Dr Werner brought up.
Any other thoughts around, you know, for folks listening what kinds of things they should think about, if they're thinking about developing or integrating technologies like this from a primary care clinician stand point. Did you guys hear that? Yes, I think that was for Dr. Aguilara. Sorry. I think the really the key is, what value is this adding right? And I think providers have to see value in these tools in order to use them. They have to see that they're creating a real impact that they're interested in seeing. I think, in the example of our project, I think physical activity is such a core mechanism across a variety of health outcomes. Right? We providers often try to encourage patients to be more physically active. So we took that on as a pretty basic health behavior, because we think it applies quite broadly, even though we focus on diabetes and depression, we think that it can have broader implications for most health conditions. as I think, starting with
something that folks are already doing, and trying to find a way to do it a little bit better with less effort. So that we know that the conversation around,