Thank you for being here on Friday afternoon. And I want to thank the speakers for presenting their very interesting work. This is the non-CAME paper session. So let me begin. We have Dr. Yaramir Mekel, who would be presenting the data from Purdue Pharma market scan data on patient characteristics, prescription pattern and alcohol use relapse among patients with insomnia associated with alcohol cessation and the market scan database. We have Celeste Wong, who would be presenting the work from Verily Life Sciences and 115 group and analysis of hybrid care, which all of us have been experiencing the telehealth versus standard care among patients with substance use disorders. And then lastly, we have alchemist, Teresa Kauf presenting the data on treatment patterns and healthcare resources use among patients initiating medications for incident alcohol dependence. And it's an analysis from the Veterans Health Administration's a very large database. With that, welcome to our speakers and welcome to paper session number two. So Dr. Mekel, would you- Thank you very much. Sure, sure. Please. So thank you very much for giving me the opportunity to present some of our most recent research project utilizing the IBM market scan database. You know, now this thing is freezing. What is happening? My apologies, but- Sounds like a- Yeah. Sounds like a technology. Okay, now it's fine. Okay. So for the disclosure purposes, I'm an employee of Fertifarmer and I will not be discussing any products or services during this presentation. The learning objectives, as you can see on the slide, is to recognize that early recovery from alcohol dependence is associated with high rates of insomnia. It is to describe the non-interventional retrospective cohort study of patients diagnosed with AOD or alcohol use disorder that are also in remission, which we call the index state, and a diagnosis of either evidence of insomnia within six months of the index date. So it's before and after. And then lastly, I will be discussing results of this comparative study that is assessing alcohol relapse incidents in patients with AOD that are in remission. And also we're diagnosed with insomnia and compared to groups, those that are being treated for insomnia versus those that are not. And this is obviously using the IBM market scan of commercially insured population. So just as I already mentioned, I mean, literature, there are various estimates that early recovery from alcohol use disorder is associated with high rates of insomnia and ranges, and can be as high as 90% of individuals have been reported to suffer from insomnia. It also is documented in literature that sleep loss or poor sleep quality has been associated with increased risk of alcohol use relapse. However, most evidence has been gathered from small trials and really not much has been done in a real world setting. And to examine, you know, in this world setting to examine the impact of insomnia treatment on AOD relapse. And that's precisely what we set out to do. So as far as the study is concerned, the primary objective is to assess incidence rate of alcohol use relapse in patients with AOD in remission who had received, who were treated for insomnia versus those that were not. This is a, or was a non-interventional retrospective cohort of patients with AOD with, in remission, with evidence of insomnia within six months of the index date. And as I mentioned earlier, data sources, the IBM market scan commercial and Medicare supplement databases. And we had used the years of 2008 through 2019. Here's a table that provides you with some demographics and clinical characteristics of the population we had studied. And as you can see, many of these patients were treated. And so we've separated these data into those that were treated versus untreated since these are the comparatives. And 35,000 of these patients were treated versus about 1,700 that were not. This population tends to be on the older side. The average age is about 49 in both groups. The gender distribution, among the untreated are more likely to be males, about 66% when compared to those that were treated, about 50-50, about 55%. But I think what I'd like to highlight here is the comorbidities, which is the rest of the part of the table. And to realize that this population does not suffer just simply from alcohol use. They suffer from other comorbidities. If you look at this table, it's quite surprising, or maybe not surprising, but some of these, the prevalence of some of these comorbidities are quite high. So mental health, such as depression, anxiety, et cetera, is very high among both groups. 71% among those that are treated, 62% among the untreated. Other substance use disorder outside of alcohol is also high in both treated groups and untreated groups. We have a third of the population that suffered from this. And I'm not gonna go through individual, but I'd like to just point out the last row, the cardiovascular disease. So we have half the population we identified in these years in this database actually suffered from cardiovascular disease. And that includes hypertension and those kinds of things. I mean, I'd like to go more into detail, but I have 10 minutes, so I kind of have to run through this quickly. So what have we done in terms of analyses? We have split the groups into three different parts. We tend, mostly because they differed. And so we want to make sure that the signal, if there is one that we detected, is consistent across the different groups. So we have identified patients that had entered remission without insomnia and developed insomnia post remission. And that's the first group called nuanced insomnia. Then we had the group that entered remission with insomnia and we call that the continuous insomnia group. And then finally, we just combined the cohorts and together pulled data and looked at the, again, about the incidence rates. And so there's a difference between relapse to alcohol use. So on the right side, what we had used was the incidence rate ratios. And on the right side, it's graphically presented. And you can see the horizontal line there. It goes from one to above one to less than one. And just to orient you above one means increased risk for relapse among those that are treated. On the left side, it is displaying those that treated patients for insomnia are at low risk. So what did we find? Unadjusted data. We have found that on both cohorts, including pooled, we have found a strong signal that treating insomnia is reducing the likelihood of relapse to alcohol use roughly between 25 to 28%. So it's about 20% on the nuanced group. It's about 25% reduction in the continuous insomnia patient population. And overall, about 20%. The next step that we did, we said, okay, well, that's fine. That's unadjusted, that's great. So as normally what epidemiologists do is to adjust all of our data. And we chose a methodology propensity scoring, which I'm not gonna go into details, but basically it is to make sure that we account for potential imbalances between the groups or any potential confounding. And then we used a so-called inverse probability of treatment weight. And with that analysis, we have found very similar results. And so among treated patients, we found, if you look at the nuanced, about 27% reduction compared to those that were untreated. Among the continuous insomnia group, we found a 19% reduction compared to those that were untreated. And overall, about 21% less reduction in relapse, alcohol use relapse among those that were treated compared to those that were not treated. So we felt very comfortable and excited about these results. So as any study, we do have limitations. And I don't wanna read all of these, but certainly they're listed here. So as far as the study limitations, as any claims data, the IPA market scan may not generalize to all populations in the U.S. And that includes those that are uninsured, underinsured, uninsured, and that may not have adequate access to healthcare. I'm just, you know what? I never stop my clock, so I have no idea where I am in the time, but I think I'm on. You know, because this is a commercial database, it includes folks that are employed, covered insurance, or their dependents that have insurance. Further, the diagnostic codes that we used for the AUD and remission, that we used to identify patients and so just by design, this analysis may be limited to only that subset of the population that suffer from AUD that actually seek treatment for AUD. The other point is that analysis was also relied on the presence of prescription drugs claims for an on-label and off-label insomnia treatments to identify the treated insomnia population. So, and what we had used there is a methodology that was just recently published by Dr. Amari and her group, and there's a reference below. And then lastly, this database does not include any over-the-counter treatments, does not include any herbal remedies, does not include cognitive behavioral treatments or any other non-prescription insomnia. So, these are prescription medical treatments. So, I'm not sure, maybe we presently be doing more work on this and see if we can gain any data, which is quite difficult, but at this point, this is one of the limitations. So, in a summary conclusion, for patients with AUD and remission with and without a prior history of insomnia, alcohol use relapse was significantly lower in those who received prescription medical therapy for insomnia versus those who did not receive prescription treatment in commercially insured population. And that concludes my presentation. I know it was quick and I just ran through that, but that's by design. And I would like to thank you for your attention and I'm looking forward to the Q&A session later on. Thank you. Hi, everyone. My name is Celeste. Today, I'm gonna be sharing analysis looking at the impact of telehealth within a hybrid care model on treatment retention and engagement in patients with substance use disorder. I have no relevant disclosures. Educational objectives for this presentation include discussing the benefits of a hybrid telehealth model of care delivery, identifying challenges to service delivery during a pandemic event for patients treated for substance use disorder, and interpreting results of analyses of patient engagement, looking at comparisons between in-person, remote, and hybrid care delivery. So briefly, as a little bit of background, as we're all aware, we had a pandemic that resulted in shelter-in-place orders throughout the US. During this time, telehealth became a very common method of delivering care, including for substance use disorder. And even though telehealth resulted improved access prior to and during the pandemic, unfortunately, there is little research that exists on its use in the substance use disorder population in particular, as well as how telehealth may impact engagement and retention in the substance use population. And so we wanted to test our hypothesis that our treatment program's hybrid care model is related to greater engagement and retention in care. Oh, geez. And so the setting of our study is 115, which is located in Dayton, Ohio, which offers a comprehensive spectrum of services. When the shelter-in-place orders were initially issued, we quickly deployed telehealth-based workflows to offer a hybrid care model where patients could access their services either remotely by phone or by telehealth with video or in-person. And so we conducted a retrospective cohort study using EHR data among approximately 1,000 adult patients admitted into our treatment program. Inclusion criteria included being diagnosed with at least one substance use disorder and having at least one additional treatment visit after the initial diagnostic evaluation. And in order to ensure that our results were not biased by patients who had multiple episodes of care, we only included each patient's most recent episode of care in this study. So we divided up our sample into three groups. One is the telehealth group. These are patients who had at least one remote telehealth with video visit. We have the telephone group. These are patients who had at least one telephone visit but no telehealth. And then we have the primarily in-person patients. We assessed survival curves of time to drop out using the Kaplan-Meier method. And then we also conducted Cox proportional hazards models to assess time to treatment dropout. And patients were considered to be dropped out if they did not have any follow-up visits for at least 90 days and they were censored or excluded from the at-risk denominator if they were still in treatment at the end of the analysis period. So here are the characteristics of our patients. Overall, about a third of our patients were in the telehealth group. About 50% were in the telephone group and the remaining approximately 20% were in the in-person group. So overall age at intake was 33 years but compared to both telehealth and telephone, the in-person patients were significantly older with a median age of 37. While the telephone group, they were more likely to be non-white, the in-person group was more likely to be white. The in-person group was also more likely to report being homeless and unemployed compared to the other two groups. And then a pretty even distribution of education. When we look at primary substance use diagnosed at intake and the overall population, a little over a fourth have a primary alcohol use disorder, about another fourth have a primary opioid use disorder, a little over a third have a primary cannabis use disorder and the remaining approximately 14% have an other substance use disorder. But when we compare across the groups, the in-person group is significantly more likely to have opioid use disorder. The in-person group is also more likely to be self-referred into the program, whereas the telephone group is more likely to be self-referred from the criminal justice system. And then finally, the in-person group is also more likely to have received medical services and to have had multiple episodes of care. So this chart on the left shows the Kaplan-Meier curves of time to drop out by group, where the blue curve represents the patients who utilize telehealth, the green curve represents the patients in the telephone group, and the orange curve represents the in-person patients. On the right, we also included dropout rates for each of the groups at various time points. So as you can see on the graph, where the x-axis is the days in treatment and the y-axis is the cumulative retention rates, there is a pretty clear separation between the telehealth curve and the other curves, indicating that patients who utilize telehealth are less likely to drop out of care. You'll also notice that there is a very sharp early dropout rates in the in-person group. And even though early dropout is fairly common in the general substance use population, this indicates that telehealth either by video or by telephone helps with that early engagement. And then finally, you also note that the in-person and telephone curves cross at approximately 100 days. And this is a clear indication that the proportional hazards assumption was not met, meaning that the hazard ratio varies significantly over time. And so because the proportional hazard assumption was not met, we decided to conduct two essentially subgroup analyses where we only included telehealth and in-person in the model and then separately telehealth and telephone in the model. So here are the results of the Cox model where the telehealth and in-person groups were included. On the X-axis are all of the predictors that were included in the model. And then the Y, excuse me, Y-axis are the predictors. And then the X-axis is the hazard ratio. One indicates there's no relationship between the predictors and treatment dropouts. And anything less than one indicates a protective factor. Anything greater than one indicates a risk factor. So as you can see at the very top, when we compare to in-person patients, patients who utilize telehealth have significantly lower hazard of dropping out after adjusting for these demographic and clinical characteristics. Being unemployed and homeless are associated with a higher hazard of dropping out. Interestingly, being referred from the criminal justice compared to being a self-referred patient is associated with a lower hazard of dropout. And then finally, having received medical services is also a protective factor for dropping out. I did not include the results of the model that include telehealth and telephone given time restraints, but the results are similar. We see that compared to the patients in the telephone group, patients in the telehealth group were significantly less likely to drop out after controlling for all of these same covariates. I didn't include a limitation slide, but one of the main limitations is that this is an exploratory and observational study. And so we cannot make any confirmatory conclusions. So more studies are needed to confirm that telehealth is just as or more effective than in-person care. But overall, our study found that patients who use telehealth specifically with synchronous video connection were less likely to drop out of care and remain in treatment compared to patients who use telephone services or primarily in-person services for the visits. In addition, our study found that vulnerable populations, including those who are older, homeless, unemployed, they were less likely to use telehealth and independently, they were also more likely to drop out of care. So this suggests that additional interventions may be needed to support these populations. And then finally, as I mentioned, we can't determine causality, but our findings do support the use of telehealth in substance use treatment models and advocate for relaxation of the rules and regulations that govern telehealth for substance use treatments. Our findings also support a growing body of research suggesting that telehealth may yield positive treatment outcomes. And some of the hypothesized mechanisms include that telehealth may improve access to care, decrease barriers, reduce costs, circumvent stigma, and provide greater flexibility for both patients and staff. And that's it. Thank you so much. So my name's Teresa Kalf, and on behalf of my colleagues that you see here, I'm really happy to welcome you to this presentation of our study of treatment patterns in healthcare resource use among US veterans initiating medications for alcohol dependence. This study was funded by Alkermes, which is the manufacturer of extended release naltrexone, one of the study drugs examined. In this presentation, I'll highlight the substantial comorbidity burden among the study population of US veterans with alcohol dependence and describe the patient journey and patterns of healthcare resource use resulting from the initiation of medical treatment for AD. Veterans are at high risk for substance use disorder, including alcohol dependence, and VA guidelines recommend medical treatment with one of the four FDA approved medications for alcohol use disorder, or MAUDs, for patients with alcohol dependence. And in fact, the availability of all approved MAUDs within the VHA system is mandated. So while veterans have access to medical treatment for AD, there's not much literature comparing outcomes across MAUDs in this population. So that's what we set out to do in this study. And we focused on the examination of treatment patterns and healthcare resource use as important indicators and outcomes of treatment with MAUDs. So this was a retrospective study using the VA's corporate data warehouse, which contains information on healthcare encounters that take place at VHA facilities. So figure one on the right depicts the study design. We included veterans 18 years of age and older who initiated a MAUD for new or incident AD. Because extended release naltrexone also is indicated for opioid dependence, we excluded patients with a diagnosis of OUD prior to initiating one of the four study drugs of interest here. The MAUD initiation date served as the index date with the 12 months pre and post index date periods serving as baseline and follow-up respectively. Veterans and others with VA benefits may or may not use VHA services for their medical care. So to identify patients with regular use of the VA, individuals with less than two encounters during the baseline period were excluded. This was primarily a descriptive study to examine patterns of MAUD use and compare healthcare resource utilization in the 12 months pre and post initiation of MAUD treatment with no formal hypothesis testing. Before we get to the primary results, I wanna take you through the derivation of the study sample. So as you can see at the top of the figure, the data set included about 600,000 veterans with at least one encounter that included an AD diagnosis. And about 12% of those patients filled a prescription or received an MAUD. The box on the right depicts the exclusion criteria. And after applying those, our sample comprised about 31,000 individuals who received an MAUD for new or incident AD, the vast majority of whom received oral naltrexone. The table on the left shows some of the baseline characteristics of the study population, which was predominantly white, male, not employed, and in their late 40s, early 50s. And here you can start to get a sense of how this group of veterans is different from the general US population. Particularly notable among the population was the high proportion of mental health comorbidity, such as major depressive disorder and PTSD, similar to the population that Yarmier presented. These patients also were frequently diagnosed with other non-opioid use disorders, substance use disorders, again, similar. We saw these patterns across all four MAUD cohorts. There were some differences between patients who received extended release naltrexone compared to the other three agents, as you can see here. The table on this slide reports a number of indicators and measures of MAUD treatment. At the top of the table, you can see the times from initial AD diagnosis to MAUD initiation for each of the four MAUDs in the study. The mean times range from about seven and a half to nine months, and you can see that the median times were shorter and in the two to three-month window. Adherence in this study was measured as the proportion of days covered over the 12-month follow-up period. The PDC is similar to the medication possession ratio, or MPR, except that PDC excludes overlapping day supply so that adherence cannot exceed 100%. We observed that the mean PDC was significantly higher for extended release naltrexone compared to the other three MAUDs. Similarly, the time to treatment discontinuation, which was defined as a 45-day gap from the previous record, was significantly longer for extended release naltrexone by about a month. More than half the patients discontinued their initial MAUD treatment, including most of those receiving extended release naltrexone, and some of what appears to be concurrent use, as shown in the bottom of the table, could be a switch in therapy or bridge therapy until a particular MAUD is available. So about half the patients who discontinued their indexed MAUD initiated subsequent therapy within the 12-month follow-up period. The chart in this slide shows that across all the cohorts, the most frequent subsequent treatment was reinitiation of the index treatment. So it's not shown here, the median time to reinitiation of index treatment when it occurred generally happened within about two to five weeks, which could reflect more of a gap in adherence rather than a true treatment discontinuation and then start of a new treatment episode. This slide shows the proportion of VHA patients with healthcare resource utilization during the 12-month baseline and follow-up periods by resource category and indexed MAUD. So moving clockwise from the top left, the quadrants depict the proportion of patients utilizing inpatient, outpatient, residential treatment center, and emergency department care in the baseline and follow-up periods. So while the levels of use varied a little bit by MAUD, you can see the extensive use of healthcare in the baseline period, reflecting that high comorbidity burden of the population. Resource use remains high in the 12 months following MAUD initiation, but with a notable shift from inpatient to outpatient care, including RTC visits. So within that RTC block, you have inpatient and outpatient. So you see the inpatient go down and the outpatient go up. This slide is similar to the previous one, except that it reports the numbers of visits as opposed to the proportion of patients with a visit with similar patterns. So regardless of MAUD, outpatient use increased during the follow-up period and inpatient use decreased. Emergency department use both as a percentage and numerically was essentially flat. So overall, we observed that about 12% of veterans with a diagnosis of AD received treatment with an MAUD from 2014 to 2018, which is higher than previous reports. Veterans have access to all FDA approved MAUDs and VA treatment guidelines suggest the use of extended release naltrexone when adherence is a concern. So that recommendation appears to be borne out in this study, which showed longer time to treatment discontinuation and longer time to treatment discontinuation and better adherence with extended release naltrexone versus the other MAUDs. Importantly, this study did not control for baseline differences in patients receiving extended release naltrexone versus oral naltrexone, acamprosate, or disulfiram. And it's possible that those differences could explain the improved adherence that we observed. Similarly, we did not consider whether the decreases in inpatient use and increased outpatient use that were observed varied by index MAUD. We note only that these shifts occurred for all the MAUDs in the study. While there was no pre-specified hypothesis testing, the nominal P values based on the comparisons may be a value for prioritizing hypotheses for future research. So this is one of the first real world studies to examine treatment patterns and healthcare resource use among US veterans initiating medication for alcohol dependence. We observed improved adherence and time on treatment for patients with new or incident AD who initiated treatment with extended release naltrexone compared to the other MAUDs. And for all MAUDs, we observed a shift from resource intensive inpatient care to outpatient care, potentially reducing the cost of care and strengthening patient engagement, an important prognostic indicator for improved long-term outcomes. So thank you for your interest in the study, and I'm happy to take questions if we have time for that. I'll turn it over to you, Meeta. Thank you. I am going to risk turning my video on and see if that would still continue to support the technology. But thank you, all three of you had wonderful, wonderful things to share. I think we had one question already answered by Dr. Mikkel to the person who had the question. We have two questions, and I think one is from, one is for, I think they both are for Celeste. So Celeste, if you can turn your video on and take the questions. The first question is from Ryan. Ryan is asking, were homeless patients less likely to receive medical services? Yes, I didn't have that in the slides, but yes, they were less likely to receive medical services. All right, and the second question for you is from Vinod Rao. And what he is referring to is because the telehealth group was the most inclusive group, having a video would rule them out of being in the in-person group or in the phone group. So it would seem like someone in one of the other group would either stay in the treatment, drop out, or become a telehealth patient. But the telehealth group patients could either drop out or stay in treatment because they can't join the other cohort. It would seem that would create a bias that would make telehealth more likely to retain in treatment. I would be curious if you reran the analysis, redefined the cohort so that the people who used phone plus video were defined as a phone. If you would have the results, could you, you could do the same analyzing, defining people who attend in-person and video. So Vinod, if you wanna turn on your video and also participate in the discussion with Celeste, I think it's more like a comment, but I don't know, Celeste, if you have something that you can reply back to Dr. Rao. Yeah, thank you so much for your comment. One thing I will say is that to your comment about how telehealth group patients could, they could drop out and then they couldn't come to the other cohorts, we wanted to minimize that bias. So we only included patients in their most recent episode of care. And we used that to define which group. They went into. Dr. Rao, are you still on this presentation? Would you be able to turn your video on and participate in the discussion? We may have lost him. The last question is, okay. All right, sorry. He's indicating that he cannot have. Yeah, yeah, I'm just looking at it. So do you have, Dr. Rao, do you? So he has a comment, but people can have multiple types of visit in an episode. Yes, that's correct. They can have multiple types of visits. The telehealth group are those who have at least one telehealth visit. And then the telephone are those who don't have any telehealth visits, but a telephone visit. They could also have in-person visits. The in-person group are those who have primarily in-person visits. I think like a very small proportion had some telephone-based services, but no telehealth. Thank you. I thought I saw one, another question pop out, but then it disappeared. So Ryan must have decided. I just typed an answer to it. You answered to him, perfect. Thank you. Dr. Mikkel, I had a question for you. So I saw that you had excluded all the over-the-counter and other non-formulary or non-FDA approved medications. So would you be able to share what particular medications were used to treat insomnia that you had included in your cohort? Yes. So it's not that we excluded, we just had no access to that data, right? I got you, got you. The claims that it does not include it. But if you look at my presentation, I actually have a slide that includes all of them. Maybe I- If I can share. I was paying too much attention to the question, the chat and the question. Let me share and I will show you that. So there was a, let's see here. So if we go to conclusions. So after that, there are some backup slides that I figured I'll include for those that are interested. And one of them is the slide right here. So, okay. Okay. Okay. Excellent. Okay. So they are the whole spectrum. Yes, yes. Okay. Excellent. So let me just double check last one time if there are any other unanswered questions. Okay. I think Dr. Rao has something more to say. If in an episode of treatment, they have a telephone appointment and a video appointment that would be classified as a telehealth group, it seems like this asymmetry allows for the possibility of a bias. Maybe this doesn't turn into a measurable bias. Celeste? Yeah, absolutely. There, this could be a potential unmeasurable bias. One of our limitations of this study is that we're unable to confirm the assumption, first of all, that all patients were given the option to participate remotely. We know that both patients and staff have varying degrees of experience and comfort in using telehealth. And so some patients may have been encouraged by their providers to use telehealth at varying rates. And that could produce some potential bias. We also don't know whether those in-person patients who dropped out early or really from any group, if they would have converted to one of the other groups if they had remained in care longer. But yeah, as I mentioned, this is an exploratory observational study. I really appreciate your feedback and your questions. And we'll look into doing, to do some further analysis. Thank you. I think we are way over our time limit. Thank you all for participating. And all three of you had some wonderful, wonderful data to share. Exciting for future directions. And thank you again.

insomnia

alcohol relapse

telehealth

treatment retention

substance use disorders

extended-release naltrexone

FDA-approved medications

veterans