So I'm Ed Bernanke, an ACOM fellow and past president of ACOM. Today we're having a presentation by our little research group that we've had going for many, many years. And this is session 234, attributes associated with high workers' compensation claim costs and delayed return to work. And really information that can be utilized by occupational medicine practitioners. And I think the other part is, hopefully, is the methodologic work that we do to develop these hypotheses and do the research. And I think being part of this group for 20 years, that to me was the key in really the whole group learning about the process. So a bottom line is, we started this in 2004, and it was really the brainchild of the director or VP of research at Louisiana Workers' Comp Corporation in Louisiana. And he invited Dr. Tao and myself to, he thought that using epidemiologic approach to analyzing workers' comp claims might be more beneficial than an actuarial approach, where really you're looking at the whole claim file and not really building up costs and time lost from work on an individual claim basis. So we've learned so much from these insurance carriers doing the research over the years. They taught us more than we taught them. So we started out in our little research group in Johns Hopkins and a professor from the University of Maryland who's with us today. And then we morphed into a much larger group of getting medical directors from the accident fund group to be part of the group, Texas Mutual Insurance Company, and their research directors at these companies to really enrich the whole dialogue. And so we have practicing physicians, occupational physicians who are more population health oriented, and epidemiologists in our group to create the hypotheses and test what's going on in real life situations where we can really use this information to predict claim costs. So Dr. Tao has used all the work that we've done all these years to create an analytic tool where we can actually predict how long a person's going to be out from work and the cost of a claim at various points in the claim process. So we think that is an interesting innovation. So today we have four speakers, and they're all from our research group. And the first speaker is Dan Hunt, who's the corporate medical director for accident fund group, which is a national workers' comp insurance carrier. And he's been in this role for the last 10 years. And prior to this time, he was a general surgeon for many years. And his interest is in pain management of injured workers, behavioral health and workers' compensation, and the impact of changing medical technologies on the surgical care of injured workers. Dr. Hunt. Thanks, Ed. Welcome, everyone. Thanks for coming in to listen to us as we kind of report out on what we've been doing for the past year in regards to our research. I thought it might be a little helpful to set some context for you and give you some background to give you some kind of a framework in which to put the paper I have the privilege to present. And that's our paper that looked at duration of claim during the COVID pandemic as kind of reflected in the claims data from AF group. This is our slide that says none of the presenters have any interest to disclose about our presentations. And so in regards to the background, you know, I just find it amazing, you know, how does an osteopathic surgeon from mid-Michigan end up with this really prestigious group of researchers for the past 10 years or so working on studying different parts of the workers' compensation. And for us, one of the areas that we found interesting was during the COVID pandemic and looking at our COVID claims and trying to get an understanding about, you know, what was happening within the pandemic, what was happening within the injured workers that were covered by our insurance company. And I think to kind of set that context, I'd like to go back because this sort of sets where our data starts. And if you think, you know, the pandemic, how did it start? It actually started, you know, in the first quarter of 2020. And you may remember, you know, where you were when it first sort of came to your attention that this was going to be a problem. But I tell you, for me, it was very interesting. So it was February. My wife and I ski, and it was a beautiful sunny day, and we were skiing. And those of you skiing know that sometimes you get on the chairlift with other people. My wife's very gregarious. She always talks with them. And we got on the chair with this young woman, and we started talking with her, and we asked her about her. And, you know, where are you from, and what do you do? And she told us she was a graduate student at the University of Michigan, and she was here studying. And when we asked, well, where are you from, she said, well, I'm from Wuhan, China. And at that time, you know, I knew about COVID-19 and the SARS, you know, virus, and I knew it came from Wuhan. And I have to say, I reflexively moved an inch or so away in the chair after she told me that. But she assured me, no, no, don't worry. I haven't been back there in a couple of years, but my parents still live there. And you know, a chairlift, you have about 10 minutes to do this interview, typically. So we asked a few more questions. But what struck me was, we asked her, well, what's it been like with your parents, right? Because at that time here in the U.S., COVID was not really here to any great extent, but it was very much there in Wuhan. And she said, well, they live in an apartment building in downtown Wuhan, and they're confined to their apartment. And one hour a week, they're allowed to go downstairs to the grocery store and get the groceries they need, and then they have to go back to their apartment. And I said, well, how long have they had to do that? And she said, oh, it's been six months or eight months or so they've been having to do that. And I thought, well, isn't that really hard on them? And she said, no, no, they're used to it. Because the last SARS pandemic, they had to do the same thing for a year. And I remember thinking, and I talked to my wife afterwards, you know, I don't see the citizens of the United States tolerating that kind of infringement on their personal rights to travel and do things, right? And you're all laughing because it's, you know, you think about it, and you know, every state was a little bit different, but I'm from Michigan. And, you know, we had a fairly severe lockdown, because as you may recall, there was a really bad outbreak of COVID in the Detroit area early on in the pandemic in 2020. And so we had some pretty strong restrictions on what we could do. But that was really kind of the start of it for me about the pandemic and how important it was going to be, and kind of maybe started some of our interest. You know, the way this research group works is a lot of people here are going to talk with you today, and a few of the people are not here. But it's a group that we meet every Tuesday, so it's a pretty big commitment. We meet for an hour, and we work on different projects. And if you look at the papers that have been published, there's a broad range of interest. But starting with the COVID pandemic, we thought that it might be very interesting to look at that pandemic and kind of see what kind of insights we could gather. And as part of that, we've published three different papers. And each one of those papers tried to look at a different question that seemed pertinent during that stage of the pandemic. And so that's kind of where we're at. And the three papers, the first one was we wanted to look at and see, well, what kind of industry groups are most of our COVID claims coming from? And so a little bit of background on AF Group. AF Group, we're a workers' comp insurance carrier. We're nationwide, but we're focused in about 36 or 38 states, a very broad geography of distribution from the east coast to the west coast, from north to south. And so we thought that this data set would be something that might be very interesting. But again, understand these are only claims that came through our insurance organization for those employers that we covered. So we are not representative, if you will, of the national workforce. There's some areas in our data that we don't really have much representation. But we did fortunately or unfortunately have a big book of business in health, hospitals, extended care facilities, clinics. And so initially on, we had a lot of claims that came to us. So the claims that we're going to talk about in this study came from over 30 states of geographic distribution and kind of some of the background we already talked about. And so the methods, and this is where Grant Tao is just invaluable. Grant sort of runs our group for us and keeps us moving forward and helps us with the methods. And so the data set for this particular study was over 19,000 COVID claims that were accepted. So not just filed, but accepted by AF Group. They started in January 1st of 2020. That was the first month that we received a few claims. And we ran the claims that we put into our cohort through December 31st of 2021. We had more claims that we could have looked at, but we thought that we wanted to look at duration, right? We wanted to answer the question with this study. Is there something unique about the duration of claims over the life of the pandemic? And so we had to close the aggregation, if you will, in December, because we started thinking about doing this in the late summer of 2022. And so we wanted to have at least five months of development to kind of give us a fair evaluation. So the data was valued as of May 31st of 2022. So 19,000 claims from the beginning of the pandemic until the end of 2021. And then another five months of development, so we could get a better understanding of how duration might evolve. And we had to pick up some definitions for our claims categories. And those of you that work in insurance or work and have some familiarity with it, you know, most of the claims are usually what we call med-only claims, meaning a claim that only had medical that had to be paid, but did not have any indemnity, meaning that no salary had to be reimbursed. And then we have indemnity claims, which mean that they lasted a bit longer, that we not only had medical that we had to pay, but we also had to pay some salary reimbursement while the individual was off work. But for this group, we also included incident only. And generally speaking, incident only in the insurance lingo is the employer just notifying you, well, we had this potential claim. It may come to nothing. We're not filing a claim right now. We just want to make you aware of it. But in the context of this study, what incident only mean was these were claims that were accepted, if you will, because it was a worker who had contracted COVID that was proven either by a COVID test or a provider that said, well, the symptoms are so classic for COVID, even though the testing either is not available or maybe did not come back positive, the provider still felt that they had COVID-19. And so the incident only claims are real claims in that sense. A lot of these people did miss time off work. But those are the three claims categories that we looked at when we did our investigation. And this is a graph that kind of shows that distribution over time. I appreciate the font's a little small for those of you in the back of the room. But it's aggregated by quarters. And it starts with the first quarter of 2020. And it goes through the last quarter of 2021. And you can kind of see the bar graphs there. Indemnity is the one on top in yellow. Med only is the orange. And then these incident onlys are the blue ones below. And you kind of see that the number of claims in the first quarter were not very high. But then as the COVID pandemic started to really get going in the country, our claims started to rise. And I think probably to put it in some kind of a variant perspective, probably that was alpha. And then in the summer of 2020, we had a little bit of a lull. But then as Delta came into being, we got this huge uptick in claims that occurred towards the end of 2020 and then into 2021. And then as you may think about it, if you remember that summer of 2021, I remember it well because my daughter got married. And trying to plan a wedding during a pandemic was not easy. But there's a bit of a lull. And some of the restrictions got reduced. But then as Omicron started to take hold, we started to see another uptick in the number of claims that came into us and that were accepted by our group. And this is a slide I think that really is important for the study that we did. And it looks at the average lost time days or the duration. And the reason duration is important is, you know, we have played with the idea of trying to get some understanding of long COVID and what that meant within our claims population. But it's really hard, you know, if you look at the claims data that we have to evaluate, we really didn't have a lot of comorbidities. We didn't have a lot of easily identifiable medical problems. And so we decided that we would take duration as sort of a surrogate for the severity of the infection that the individual had. And if you see here in the first quarter of 2020, we had a lot of really sick people. And, you know, if you think about, you know, having seen that as the medical director for a nationwide insurance carrier, I have to tell you, I was a little bit nervous. And, you know, we actually had a couple of providers in my community that died. These were not people that you would have thought were at high risk of contracting COVID and having a serious illness, but they did. And so the durations were really, really long during that first part of it. The other thing that we did is we thought it was important that we have kind of a uniform claims approach, especially if we're going to be looking at claims experience over a period of time. And early on in the pandemic, you might recall, you know, infectious diseases are not typically thought of as a workers' compensation compensable problem. But COVID was certainly a particular problem that was something I've never seen in my lifetime, and I suspect perhaps not in your lifetime either. And so we felt that we shouldn't really leave it just to the claims handler, if you will, who has, you know, have a variety of experience when they come into that position to decide whether a claim should be accepted or not. And so we put together a committee that I was on it, and we had leadership from all the claims departments. And we had, you know, the legal representation. You might say, what do you need lawyers for? What we needed lawyers for was a lot of states started passing presumption laws. And those presumption laws, we had to be very careful that we stay within those guidelines within each of those jurisdictions. And so we met, and we reviewed every claim. And we sometimes had to meet for two or three times a week, and that had to be for two or three hours. It was really arduous, but we wanted to be fair in making our determinations about the injured workers. And we followed that through. And kind of the way I knew that the pandemic was getting better is we didn't have to meet quite so often. And then as we got into later in 2021, we didn't really meet very often at all. And that's because, to my experience anyway, we didn't have many deaths. We didn't have many hospitalizations. We didn't have very many severely ill people. And so I think that kind of is what got our interest going. And well, what did we really see in our claims data? So the average duration, you can see it was 77 days in the first quarter, but then rapidly started to diminish. And then looking at it a little bit different way, we decided to divide it into three different groups. And the long-term, so 30 days or less, 60 days, or the really long-term claims, 150 days or more. And those are represented by those various graphs there. But they follow the same kind of a curve, if you will, that we had a lot of very long-term claims early on in the pandemic, but that rapidly started to diminish. And then if you look at average paid and certainly those of us in the insurance industry, we know that claim cost, ultimate claim cost, meaning indemnity plus medical, is driven 80% to 90% by the duration of the claim. The longer the claim stays open, it's sometimes a reflection of severity of illness, that's typical. But it also means that sometimes you're having some challenges with getting that person back to work. So claim duration usually always coincides with claim cost. And this was a graph that shows our average paid on the indemnity claims by quarter. Again, following the same kind of a curve with a rapid diminution as you got into the first quarter of 2021 and so on. And I think what's very interesting is, you know, of course, you, like me and everyone on the research committee, you probably ask a question, well, why is that? You know, what changed? Why do we have so many severe long-duration claims in the first couple of quarters? But yet, as we got into 2021, when really the pandemic was still going strong and it was a big part of our lives here in the United States, we saw that most of the claims that we were getting with people were not very ill, did not have very long duration of their claims and consequently were not very expensive claims. And so as you think about, you know, what was going on during the pandemic, vaccinations were starting to be distributed in the first quarter of 2021, became kind of more readily available by the third quarter of 2021. But we saw in our claims data that the duration of the average claim was starting to diminish, even before vaccines became readily available across the country. So I think it would be logical to suggest this probably was something else driving that reduction in severity. And so what are our takeaway messages? And this is where you have to be careful as a researcher and not claim more than what your data shows. And this is really summarizes what our data shows, that if you looked at the proportion of claims with lost time from work, that it was quite high during the first couple of quarters of the pandemic, but rapidly reduced itself, that the secondary finding was that cost followed along with that duration. And so that level of impairment that was suffered by our injured workers from around the country with a broad variety of industry types, really started to diminish pretty quickly. And I think that's, you know, really great news and maybe a lesson to keep in mind when the next pandemic comes along. And then of course, there's always a few limitations. And I think to point those things out, I would like to say that, you know, despite Grant Tao's wizardry with data and analyzing our data, we really cannot make any kind of relationship or any kind of a statement about what really drove that reduction in claim duration. Was it viral mutation? Was it vaccination rates? Was it natural immunity? Whatever it might be. You can make some educated guesses based on the availability of vaccinations and things like that, but we make no claim as to what was driving that. And that the, you know, the institution of presumption laws in a lot of states across the country, which had various levels of groups that they included, but that certainly had an impact on the accepted claims. Although I would tell you the vast majority of the claims we accepted, we would have accepted whether there was a presumption law in place or not. And that there's also the possibility that the work comp database that we have didn't include every injured worker who had a COVID diagnosis. You know, these are only injured workers who filed a claim because they had significant time loss from work. Or the other cohort we saw, which probably accounts for the incident only claims was, the employers would do routine testing. That was very common in a lot of industries. Every week somebody had, you had to be tested. And those individuals that tested positive would then have to quarantine at home for a period of time. We, those are the kind of the incident only claims that we received. That was kind of what filled up that bucket for us. And the last thing I just want to leave you with, and these are sort of why I love doing this research with this group is it's fascinating to me as a physician, but it's also has applicability of business, a business applicability for our group. And that was, you know, one of the things that we were worried about as an industry was, is there going to be another shoe to drop? Are we going to start to see a lot of claims that were maybe minimally expensive with short duration in the initial part, but then have COVID symptoms that then developed further on and then had to come back and they had to be have additional time off work. We really saw very, very, very little of that. And so I think that, you know, the comforting thing about the study shows that the real severe illness was in the first several quarters of the pandemic, that we didn't see a lot of severe illness after that. And that the cost for the insurance company, which again, we don't mind costs that we anticipate because you get building a rate structure. But when you have something like this comes along that no one anticipated, it was not built on the rate structure, you know, that then has some financial implications. So it was very comforting to know that we didn't see a big increase in claims that were reopening because of COVID symptoms coming back. So thank you for your kind attention. I'm going to turn it back over to Ed. Oh, sorry. And this is my plug for our research group. These are all the people that have contributed to this paper and the other papers that we write. And so that's a picture from the paper that was published in JOEM early this year. Thanks, man. So we can take a couple of questions now or we can wait till the end. Yeah, come on. I got one real quickly. Could you repeat that? We have no idea what he said. Oh, okay. Maybe you should use the microphone. Thank you. They were two good questions. So I have two questions. The one question is initially the CDC and health departments were recommending a longer quarantine period. And whether as it went along and they lowered the current recommended quarantine time, whether that in part wasn't related to the shorter claims. And the second question, I'm from Temple University, the city of Philadelphia mandated that we cover all COVID related employees. In other words, we couldn't use their sick time. We had to basically pay them even if they had no approved sick time. And I wonder whether the laws and local jurisdictions impacted the time, the duration of the claim. Sure, as Dr. Bernanke said, two very good questions. I think part of the answer lies in that there were a lot of variables, states at different times enacted presumption laws that had certain requirements about quarantine time at home and paying the employees for their time off. And certainly if you look at the med only claims, the incident only claims, and probably some of the shorter duration indemnity claims, that might've happened, but we were hopeful that with 19,000 claims over extended period of time, that kind of those little bumps in the road, those minor changes in the data might not have a big impact. But we don't have the ability to answer that question because this was a data set from across the country. There was a lot of local variation. The other thing I think that just from my experience at the organization and looking at the claims that we manage the long duration and the high medical cost in the first quarter, those things were driven by people really sick. We had a lot of people in the hospital. We had a lot of deaths. We had some very, very ill individuals. And I think that that was so overwhelming. We still have numerous claims that are still open today that people have had persistent disabilities that are probably never gonna get better. So that I think that that was the overwhelming kind of wave that drove that data was the severity of illness that we saw during the first two or three quarters of the pandemic. Yeah. Hi, thanks very much. Bill Buckner from Wisconsin. I'm just wondering if the rapid drop in 2020 in the second, third and fourth quarters before the vaccine might be attributable to the application of the non-vaccine interventions, face coverings, physical distancing, which would reduce the viral load. And many of us think that a lower viral load, a better chance your immune system has to fight the virus. No, I think that's a great point. And I guess it's back to, we really can't draw any conclusions about what drove this change. All we can do is report out, this is what we saw in our data. And obviously there were so many things going on during the pandemic, changes in our approach, changes, improvements in treatment for people that were severely ill. You know, it's very, and I think that's probably why it's kind of an easy thing to say, well, we're not really sure what drove it, but this is what we saw. Hi, with respect to the incident only claims, for those individuals who had a bona fide positive test or even a very asymptomatic infection, when they were quarantined, who paid them? It varied from jurisdiction to jurisdiction. And we saw it all over the board. You know, if they actually, if it became a claim that they gave to us and said, we want to file a claim and they met the criteria of, they were either in a presumption state or they had a positive test and they were off for longer than whatever the period of days might be. Sometimes it's three in some states, some states it's seven. Then we would pay indemnity on that. And those went in that bucket of indemnity claims. So not all the indemnity claims are severely ill. Some of them were just people that were off for two weeks or three weeks because of quarantine requirements. Got a question from the streaming audience. How were work on COVID patients managed in regards to MMI and permanent impairment? Wow, that's a really good question that I think is beyond my abilities. I think that, you know, we tried as an organization to manage the individuals just as we would any other claim that came in. You know, it was unique because it was infectious disease, but I don't really know. I don't have the answer to that. I'm sorry about if there's been any differences in how we determine MMI or permanent disability about that. All right, well, thanks so much for the questions. Appreciate it. Well, thanks, Dan. Our next speaker is Dr. Robert Lavin. And Bob is gonna be talking about trends in combined opioid and gamma pentanoid prescribing and workers' compensation. So we're totally pivoting here. Implications for clinical practice. Bob was until recently an associate professor of neurology, physiatry, at the University of Maryland School of Medicine and is currently adjunct associate professor of medicine at Johns Hopkins University School of Medicine. Bob has treated injured workers in the Johns Hopkins Workers' Compensation Clinic for roughly 25 years. And he participates in training the occupational medicine residents at Johns Hopkins. And he engages those residents in how to make a clinical diagnosis with work-related illnesses. He oversees a multimodal management for employees with the diverse range of musculoskeletal and peripheral nerve injuries, including thoracolumbar and cervical pain syndromes and repetitive trauma injuries. Bob? Need to get to the right slides here. Okay. I'm looking over the edge. Thank you again, Ed, for that very complimentary introduction. And I want to thank you for your hard work in making this team as productive as it has been over all the years. You really deserve a great deal of credit for that. We're talking about in this talk about trends and combination of opioids and gamma pentanoid prescribing and workers' compensation and the implications for clinical practice. I have no disclosures. In the previous paper, Dr. Liu, we looked at cost trends affecting opioid and gamma pentanoid utilization. We looked at opioid and pregabalin prescribing, which decreased over that time period, and gamma pentanoid prescribing increased. So we wanted to look further at these trends and look at them in combination as they occur in the real world. So that was the impetus behind the current study. Gamma pentanoids have been approved from the FDA for treatment of post-traumatic neuralgia, diabetic peripheral neuropathy, and spinal cord neuropathic pain, but they've not been approved for non-neuropathic pain conditions such as musculoskeletal pain, which is more commonly seen, or at least what I more commonly see in the workers' comp population. Furthermore, during the opioid crisis, providers have been looking for alternatives, and many of them have latched on to gamma pentanoids. But again, unfortunately, gamma pentanoids are a poor substitute for treatment for musculoskeletal pain. They're primarily used for neuropathic pain, and most of our patients we see are musculoskeletal pain patients. The other concern is whether the benefits is worth the risk, because gamma pentanoids, when they're prescribed alone, are generally lower risk, but there's a much higher risk. In this study by Xu that I mentioned, they are double the risk for overdose and death when prescribed with opioids. There are other studies that show increased risk of hospitalization and emergency room admissions related to gamma pentanoids. There are other studies related to respiratory problems associated with gamma pentanoids when they're available with opioids. So the objective of this study is to determine the trends in combined prescribing of gamma pentan and pregabalin with opioids at different doses among Louisiana workers' compensation claimants over an 11-year period. So a total of over 18,700 injured workers filed injured claims with Louisiana Workers' Compensation Corporation between 1998 and 2008 and filled any prescriptions, and that means prescriptions for any drugs. It could be gamma pentanoids. It could be opioids. It could be other drugs over an 11-year period ending between 2008 and 2018. Louisiana Workers' Compensation Corporation is a private organization Compensation Corporation is a private workers' compensation insurance company that serves approximately a quarter of a fully insured workers' comp population in the state of Louisiana. So for the purposes of the study, we divided the claimants into 10 groups based on the prescriptions, first group being those who had no opioids, no gamma pentanoids. Another group had gamma pentanoids only. We also further stratified the opioid groups based on milligrams, morphine equivalent, daily dose, and you can see the dosages there, 1 to 19, 20 to 49, 50 to 99, and 100 or greater milligrams, morphine equivalent, daily dose, and further divided those groups between those who were non-gamma pentanoid and those who had gamma pentanoids simultaneously prescribed with the opioids. So in this graph, I want to point out several things. This is, we looked at the trend of drug utilization for these 10 groups that I just described in the previous slide by the script combination and the calendar years 2008 through 2018, and what you see highlighted in the yellow are all but one of the opioid groups had a decreased trend in prescriptions of these drugs. The only opioid group that showed an increase, which is highlighted, and those increases are highlighted in green, was one with low-dose opioids combined with gamma pentanoids. The only other groups that showed an increase were the ones with gamma pentanoids or the ones without either of the two study classes of drugs. What I also want to point out is that the high and very high-dose opioids were prescribed in relatively smaller amounts and that decreased over the time of the study, and you can see, since this is a logarithmic graph, that the actually three out of the four high, very high-dose opioid groups fell below 100%. This doesn't want to... There we go. Okay. So that was the only slide that has the logarithmic scale. The rest of them are regular percentage scale here, and what you can see in this graph is a trend of gamma pentanoid prescribing with opioids by prescription year, and here we did not stratify the opioid dose. So what you see is gamma pentanoid with opioid and... Excuse me, gamma pentan with opioids and gamma pentan alone increased over the course of the study, whereas pregabalin with opioid, and it's hard to see it, but also pregabalin alone, because it's on the lower end of the graph there, decreased throughout the course of the study. We saw this with our previous study as well, and we hypothesized that this is probably related to price, since throughout the course of the study, gabapentin was a generic, and pregabalin was a brand-name drug, and there was probably a significant financial impetus to move patients onto the gabapentin medication. The other thing I want to point out is there's two kind of unusual combinations up there, which are a combination of two anticonvulsants in the same class, pregabalin and gabapentin, and with or without opioids, and we suspect the reason for that is because those patients were being transitioned from one drug to the other, probably due to adverse drug effects, possibly due to efficacy issues, or they were being transitioned to the gabapentin from pregabalin because of cost issues, but those are all speculation. As I mentioned previously, when we get to the high and very high-dose opioids, what we're going to see is more variability, which is what you're going to see in the next two graphs. So in this figure, you see daily gabapentin dose per injured workers by opioid dosing group. So those with a higher opioid dose, a very high and high opioid dose, also had the higher gabapentin daily dose prescribing. Those with a moderate opioid daily dose had also a moderate daily dose of gabapentin prescribing, and those with no opioids or the low opioid daily dose also had low gabapentin daily dose prescribing. So there seems to be an association between dose and between the two different groups of drugs. This is pregabalin. Again, fewer of the claimants were prescribed pregabalin, and again, at high doses, you're going to see more variability, which is what we see especially toward the end of the study because there were fewer claimants on both high-dose opioids and pregabalin. But what you do see is for the claimants who were on very high-dose opioids per day, they were also on higher doses of pregabalin daily. Those who were on high and moderate doses of opioids daily were on more moderate doses of pregabalin, and those that were not on opioids or on low-dose daily opioids were on lower doses of pregabalin on a daily basis. So again, there seems to be an association between the pregabalin and opioid dose. So in summary, the proportion of claimants prescribed opioids alone decreased at all dose levels, and this is primarily due to reduction in the high and very high opioid prescribing. Gabapentin prescribing was associated with continued prescribing of opioids, and higher daily doses of gabapentin and pregabalin were associated with higher daily doses of opioids. Some limitations claims reflect only workers' compensation data from Louisiana Workers' Compensation Corporation, which may not be representative workers' comp claims from other states, since Louisiana has statutes that prevent workers from returning to work until their claims have been resolved. We don't know whether the prescribed drug dosages and drug combinations were associated with severity of injury. Obviously, a higher injury severity may explain higher drug doses and use of combinations of analgesics. And we do not know the proportion of claimants with mental health and substance abuse or dependence diagnoses, which is associated with increased reliance on psychoactive drugs like opioids and gabapentinoids, and those individuals, of course, would be more at risk for problems with those drugs. The concern in this study is that as we see higher doses, well, the concern is we noted initially that gabapentinoids and opioids, when prescribed together, provide an increased risk of overdose and death, and we're seeing an association between higher doses of both of those drugs, which is, of course, a concern. Implications are that opioids prescribed with gabapentinoids are common practice, and medications for injured workers should be evaluated for use in combination rather than individually to detect potential complications and avoid further risks caused by these combinations. This is our paper, and my colleagues, and I would like to thank for their assistance, and I'm open for, do we want one or two questions? Thank you. So if there's any questions, we can ask them now or wait until the end. Okay. There's not a show of hands, so. Okay. Okay. Moving on. Our next speaker is Dr. Nimisha Kalia, who's an adjunct assistant professor of medicine at Johns Hopkins University School of Medicine, and adjunct faculty at the University of Texas at Austin's Dell Medical School. She currently serves as the chief medical officer at GE. Her research interests include assessing workers' compensation costs and on-site clinic programs for employers. She has been a member of the JOEM editorial board since 2019, and she's going to talk about trends in opioid dose escalation among injured workers. Thank you, Dr. Bernacchi. Okay. So this is going to be our second opioid paper of our series, so I'll go ahead and start. I'm not sure why that's there, but, Nimisha Kalia, it's a pleasure to be here. The disclosures, the presenters don't have anything to disclose for this presentation. So the background is that opioid prescribing in general has, when you look at the literature, has been associated with a few deleterious outcomes. One is that it's associated with dose escalation, associated with prolonged prescription of opioids, and poor claims outcomes. And by poor claims outcomes, we see that the durations of claims are typically longer if opioids are involved. The closure rates are also slower if opioids are prescribed for an injured worker. And then the cost tends to be higher as well. So we've previously reported on that. Also, there's obviously causes for overdose, safety concerns from an employer's perspective, and death that is also associated with opioid prescriptions. So some studies have also documented that within the workers' comp literature, opioids are prescribed more liberally and for longer durations for occupational injuries than they are for non-work-related injuries. And for the context of this paper, in the state of Louisiana, the research that I'm going to describe is also looking at LWCC data sets. So that's also from the state of Louisiana. In general, opioids are prescribed more frequently in higher amounts for treatment of injured workers compared to other states. So the objective was to assess how and when in the course of recovering from an occupational injury, the first three months of that after injury relate to the continued use of opioids and the potential dose escalation. So essentially what we're trying to determine is the morphine equivalent dose in MEDs per day, and then look at the escalation trends over the duration of a claim. And again, for context for this particular data set, closure is defined as back to work in the state of Louisiana. Our study population, including a total of 25,108 lost-time Louisiana workers' compensation claims that were filed over a 10-year period. So the 10-year period was from 1998 to 2007. And these claims were followed for an eight-year period. So the follow-up for each claim ended from 2006 to 2015, respectively, so that it allowed for the fulfillment of the eight-year post-injury follow-up period. And then the opioid morphine equivalent doses in milligrams were calculated using established criteria. And the average MED per day for each period were calculated for different follow-up periods. So in the first year of the post-injury claim, you had the zero to three months period. Then you had the three to six months. Then you had the six to 12 months. And that's all during the first year post-injury, and then annually after the first year. And then the average morphine equivalent dose was calculated by using the total MED accumulated for the period that was specified, and then dividing that by the total days during that particular period. So for the data analysis, the workers were classified in four particular groups based on the opioid MED during the first three months. So either the injury started off with the injured worker getting no opioids, or the injured worker had one to 14 MED per day. And that was defined as low-prescription opioids. Medium-prescription opioids were defined as 15 to less than 30. And then high-dose opioids were greater than 30 MED per day for the purposes of our paper. The proportion of opioid users and high-dose users in the average daily MED were calculated at the three to six months, six to 12 months, and then annually after that. And then linear regression was used to simulate the MED escalation. So that's what we used. We used the calculation that you see at the bottom. So the daily opioid MED was alpha plus beta in the year post-injury, where alpha refers to the intercept or the simulated initial MED per day. and the beta refers to the yearly increase of the daily opioid MED among opioid users. There's a little bit of math in there. So this looks like a pretty complicated graph, but what it's essentially showing is the study period. So on the left side, you'll see the injury year. So those are the injury years that we're following these claimants from 1998 to 2007. And then as you move horizontally, what you can see is the number of claims that were still open at the beginning of the specified period post-injury. So when you take a look at the year 1998, you had a total of 2,614 claims. And then at the three-month mark, you had 2,032 claims that were still open, and that's how you read this. So you go along, and at the end of the eight-year study period for the year 1990, a claim that was opened in 1998, after the eight-year study period, you had 70 claims that were still open as you follow along the horizontal line. In total, if you move your eyes to the bottom left, there's a total of 25,108 claims I'd mentioned before. But this is, again, an accumulation of all of the different claims, and then as you follow along on the top, the claim years and the claim periods that were measured. And then the next four slides are also going to have a lot of data, but I'll try to summarize what it actually shows. So what it shows is the number and the proportion of open and closed claims among the four initial opioid groups during each of the follow-up periods. So I defined what the follow-up periods were, and I also defined what the opioid groups were. And then it also shows the progression of open versus closed claims during the respective follow-up periods. So as you move through the next few slides, what you'll notice is that the claims that had no opioids within the first three months post-injury represent 25 to 33 percent of high-dose opioid claims by the year two, meaning that even if somebody started off with no opioids, over the duration of claim development, by year two, those claims are still ending up in the high-dose opioid category. And then the initial high-dose opioids at three months do decline. So in the higher initial opioid prescribing groups, you start to see declines in opioids, but overall there's still an increased trend. So this graph represents the closure rates. So just to orient you, these are the four—I don't have a pointer, but these are the four opioid groups that we defined. The blue is the no opioid during the first three months post-injury. The yellow is the low-dose opioids. The—I don't know what color that is—green is medium-dose opioids, and then high-dose opioids as defined as greater than 30 milligrams per day. So what you're seeing is the closure rates were proportionally highest for those in the lowest opioid daily dose group, but all four closure groups had a similar type of curve. For the ones that had no opioids or low opioids, these groups, more than 50 percent of them closed it around year two, and for the ones that had high or medium-dose opioids, those—the same closure rates of greater than 50 percent took four years for them to claim. And to reach the 85 percentile, the low and medium—I'm sorry, the no and low opioid prescription groups took four years, whereas the medium and the high opioid groups took seven years on average. So the closure rates were lower in those groups. So this chart shows, again, those same four opioid—initial opioid groups, and what you see is that for the no opioid group, they do increase by 10 percent in the first six months of the post-injury, and then it starts to stabilize at around 25 percent after two years post-injury. For the low-dose opioid group, it starts to stabilize around 45 percent of the low post-injury. You do see it also start to increase at the one- to two-year mark, so there is an increase in dose escalation, but then it starts to stabilize. For the high—the medium and high-dose opioid groups, they tend to decrease, but then they stabilize as well at around the four-year timeframe to about 60 percent proportionally of individuals that are still on opioids. This graph shows the proportion of open claimants that were initially prescribed opioids at a high dose. So just to orient you again, so if you look at the no opioid group, it starts off at zero, so that's the green bar. So they start—that's the no opioid group, they were not prescribed opioids, so that starts off at zero to begin with, and then it starts to increase. And then the red, for example, on the top, that's the group that started off at high opioid groups and then started to decrease. So essentially what you're saying is how many claimants in each group had—were prescribed 30 milligrams or above of opioids, and that's what this is showing. We did start to see variability around the year four of the claims, especially for the medium and high-dose opioids, and that's likely because the number of claimants that still had open claims were much smaller proportionally to the number of claimants that had low or no opioids initially. So the variability is just because of smaller numbers after about four years. This chart looks like it has a lot of data, but just to orient you again, so when you're looking at the blue graph, that's the number of opioid users, so that's the total number of claimants that are using opioids, and you can see that it starts to decline. When you look at within the first zero to three months, it goes down to about 52 percent of the initial total claimants by the three to six months, so the first initial time period decreases the number of claimants that are using opioids considerably. And then as you break down the different opioid strengths, you can see that regardless of whichever group they started out with, the opioid dose does tend to escalate in every single one of the four opioid categories that we had selected. So among open claimants, the average daily opioid dose continues to escalate for all of the dose categories, regardless of the initial prescribing patterns. So if you were to look at open claims that had an initial dose, and I know this is a very busy slide, but if you take a look at the—if they had an initial dose of 11.3, and that would be considered low-dose opioids, if the claim is still open at year eight or nine or eight and beyond, that particular claimant would have a dose of 66.1 by the end of that claim if that claim is still open on average. So regardless of whatever group they started out with, if the claim continues to be open, there is certainly a trend that the dose escalation is going to continue. Table three shows a few different things. So one, it shows the curves for the different opioid categories that I just mentioned. And what we see is that the slope for each one of them—so if you take a look at beta, the slope varies, but it's still quite similar, and the total average slope for each one of those curves is 6.28. The other things that you see here is that the R-squared is high. So the high R-squared values suggest that 90 percent of the variability is explained by the regression model that we use, and that the low P-values also suggest that this is statistically significant of a trend. And the discussion is that annual opioid doses and morphine-equivalent doses per day increases regardless of what the initial opioid dose that was prescribed to a post-injury worker. And claimants with high-dose opioids in the first three months are associated with lower closure rates as well. And the other interesting thing that we found was that at about the two-year mark after claim development, about 25 percent to 33 percent of the open claims had high doses despite the fact that they actually started off with no opioids to begin with or low-dose opioids to begin with. And that is—I'd like to thank the rest of my research group, and thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Tha tipping point of the cost. Everyone if you were closed within the three years, the cost is very limited, but if you go beyond three years, and the cost will be increased quickly. So that's why we want to know how many of them are still open and what are risk factors that would be associated with still open at the end of the third year. The drug selection are listed, which is totally eight, and so including opioid. And the time window we use to measure or to evaluate is the first three months, which is 91 days, and the six months, the first 182 days. And the first, as you can see, the first six months included the first three months. We just tried a different time window to see which one are more sensitive. So the drug dose categories, yeah, first we converted the different type of opioids into a morphic equivalent dose in a minigram based on the cited literature and methodology. And then we stratified the MED into five categories, and zero, and one to less than five, five to less than 10, and then so on. And we do this calculation for first three months and the first six months as our potential early predictor or markers, and that is post-injury time. And for the other psychotropic drugs, because the number of users are too small, and we tried the dose as in the Dr. Levin's study, and then we found it's just too small in numbers, and then study power will get limited. So we used the ever, never during the first three months or first six months as a variable and instead of doses. But the opioid, we were using the MED. So the major talks for this paper is to have a better control for potential confounders. So first, a group of variables, a demographic, and we used sex and age group. And controlling for severities, we have three type of approach. And one is the initial reserve. The initial reserve is the amount of money and adjusted put aside within first 14 days after the reporting. And they are majorly based on the injury itself and the trying to ensure that amount of money is enough to cover this injury. So it's a basic severity control. And the second is the physical therapy. In our previous study, we found that the type of physical therapy, which is either passive or active, and the number of active physical therapy visit, those are very important risk factors associated with our final outcome. So we want to control for that. And the final severity variables we want to control is the surgery. So we classified our surgery variable into three categories, and those who never had a surgery, and those who have spinal surgery, and the other surgery procedures. And the reason we specify the spinal surgery as a separate group is because our previous studies found that they are highly associated with outcome, with a very high odds ratio. And the last one we want to control as a confounder is attorney involvement. And as we mentioned early, we are not trying to say this is causal in which direction or not causal. But we found that any claims with attorney involvement might have a higher cost and delay the return to work. And the cause of relationship might be in both directions. But that's not our focus. We just say any claim with attorney should be compared differently, I mean stratified. So that is our purpose in this study. We want to control for it. So we have two basic models. And one is the cost model. So we use the final cost equal or over $100,000 as the end point, and that's cost model. We say it's model one. The model two is the return to work model, which we use the claim still open at the end of 30 year post-injury as the end point. And so for both models, we're running two sub-models. The first model is using the first three months of drug prescription in a model. And the second sub-model is using the first six months of drug prescription pattern in the model. So totally we have three sub-models. And all of those models control for the same set of correlate we defined in the previous slide. So this became busy, and I'm sorry that there's a lot of numbers. So these next three slides is the one model. It's the model one. And the model one's sub-model one, which is first three months, and also first six months, I have these two sub-models listed in the same slide so you can compare. So this part one of the model one slide shows the different medications during the first three months and the first six months on the right. And what you can see that the opioid use in the first three months is not very good predictor. You can say that using the opioid or not in the first three months at any levels, even a higher level, higher or equal to 15 milligrams per day, it doesn't matter. It does not statistically increase the risk of the final cost over 100,000. But if you move to the second time window, which is first six months, and then anybody with the dose over five milligrams per day, and that is statistically significantly associated with the cost, higher cost, the 1.35, 1.32, and 1.74. It looks like increasing with the dose. So the opioid is the better predictor for the first six months, but not the first three months. However, pregabalin and antipsychotics and antidepressant and the sedatives, and those for a good predictor in the first three months. So they are more sensitive, and they can be used as a marker. For whatever reason, for people that use them during the first three months, it should have a flag. You need to focus on those people to see what's going on, or you have somebody, a claim manager to see what you can help. And however, when it's moved to the first six months, and pregabalin and, yeah, gabentin, and get statistically significant, 1.8 and 1.88, and the other one, too. So this shows that psychotropic medication and opioid, they can perform differently. And especially psychotropic drugs looks like more sensitive based on this model. So the second part of this model one is showing the severity confounders, which you can see that the higher the initial reserve, the higher the risk of getting the higher cost, which is higher than 100,000 at the end point. And also, the active physical therapy, or the higher number of active physical therapy visit, and the higher the risk of the higher cost. So it's increasing with the number of visit numbers, and you can see that. So that's definitely a strong confounder. And then the surgical. You can see that the spinal surgery shows, like, a 15-fold during the first three months and 13 in the first six months. Those are really very strong, the risk factors associated with our final cost. So we control it for you. And the last part of the model one is the demographic. And the female tends to have a protective, quote, association, which means that if you are female, the chance of getting 100,000 final cost are less compared to the male. And age, and it's kind of increased with the age group. And in the first three months, that higher age is not significant, perhaps because the number of claims are smaller. But basically, it's higher age will have a higher risk, and it makes sense, right? And injury year, we try to control it for whether we have differences among injury year, because we have 10 years of range. So we want to make sure it's not a significant factor. And then the attorney involvement, of course, it's six-fold, around six-fold in either first three months or six months. So those are strong risk factors. But based on the analysis, these strong confounders are controlled. And then we still found the opioid and the other drugs had a significant association with outcome, which means probably in early stage, we can use them as a predictor. So the model two is looking at whether you are still open at the end of third year, and also looking at the two different time windows, which is similar, actually, to the cost model. And you can see the opioid do not perform very well in the first three months. And even they tend to have a protective association. It's lower than one and significant, which means it doesn't matter if you use opioid during the first three months, which does not increase your risk of closure at the end of third year. This probably reflects the fact that a lot of people are using the opioid during acute period, and then once or two, and then that's it. But the first six months and the opioid use at a different level will still be a significant risk factor. So you can see that the odds ratio from 1.25 and increased to 1.38. And then precaproline and antidepressant are only two drugs, and that are statistically significantly associated with the closure are still open at the end of third year. So which means you still can use them as a marker, and just choose a choice for your model. And again, this is part two of model two, and you can still see the first reserve and physical therapy and surgery, and they are strong risk factors, which are much stronger than probably the opioid use. And also the demographic, and not very much for gender, but age. And age is a strong predictor. And also the attorney involvement, and it's also strong. And so these two models give us some kind of impression that for the first three months, we might use the antipsychotropic drugs, such as antidepressant and agavapentin as a predictor of our final cost, over $100,000, or still open at the end of third year as a predictor. But for the first six months, and then the opioid dose, the opioid level can be used as a predictor for either the cost and also not close at the end of third year. So as our colleagues mentioned before, our studies do have limitations. So that's why we are keeping on doing more research to try to fix our limitations. And for instance, so we only have the data in our claim file or PPM, which is collected after the injury, and we don't know anything on the data prior the date of injury. So and also our average dose calculation is average, which is to use the total dose of MED and divide it by the first three months or first six months, which is not cover the fluctuation, right? We don't know the peak level. We don't know the highest dose and the lowest dose. And so it's average. It's just a rough estimation of dose. And also, so there are some changes over time, which probably will influence our study. And since we cover so long, and it's 10 years. So we do have some type of limitation. So we're trying to improve in our future studies. And so yeah, the take-home message is the first three months and the first six months we can use these two type of medication as a predictor. And this paper was published last year, December last year. And you can take a look if you're interested. And thank you very much. And if you have any questions. Thanks, Dr. Tao. I think we're running out of time. So one question then. Maybe a good summary question. So I'm not aware of the current workers' comp rules in Louisiana. But did your analysis of these data over time inform their workers' comp rules in any way? Or if you did not have a chance to do that, what are the observations in terms of potentially advising a system such as that current state of their system at that time? Thank you, Craig. Yes. So works compensation database provided a source for us to do this type of analysis. Actually, if you have access to Medicare or general health insurance, that can be also a good source of data analysis. Maybe even better. And because the work time is limited and focused on the injured workers. Probably the general medicine, Medicare and medical insurance data will cover more, even more range of subject. Yeah, come here. Yeah, anyone can stay if you'd like. But I think the bottom line on all of this is that you really have to look at a workers' compensation cost for a long period of time. Because as you know, the insurer or the company or whatever it is at the time of the accident bears the risk all the way through until the claim is closed, until the employee dies. And now with Medicare set-asides, it's really a big issue. And I don't think occupational physicians realize what they can do to manage these cases for an earlier closure. And I think at least that's the appreciation I've gotten over the years. And Bob? I just want to make the point that you saw the odds ratio. And I think, correct me if I'm wrong, Grant, but they were calculated independently for each of the categories. So one of the things that helps to inform the treatment of these patients or predictions of these patients is they're multiplicative. So you can look at patients with opioids and different psychotropics and multiply those odds ratios to see a higher risk for those. Thank you. Well, thanks for your patience. I think this is the end of the day. And we really appreciated presenting. Thank you. Thank you.

workers' compensation claims

claim duration

claim costs

COVID-19 pandemic

severity of illness

prescribing opioids

gabapentinoids

Louisiana

dose of opioids

dose of gabapentinoids

injured workers

higher costs

longer claim durations