 Next we're going to hear from Brian Stagg, another of our third-year residents. Brian's going to talk to us about emergency department visits for urgent and non-urgent ocular problems. Is he all set? Yeah, great. So last presentation I gave was at Grand Rounds, and my PowerPoints didn't work on the computer, so I'm plugging in today. Okay, thank you, Adam. That really was super interesting. And thank you, Zach, for your introduction. So from Zach's introduction, I learned that I'm a nerd, but I knew that already, and I think all the residents did, too. So thank you, Zach, for helping everyone else know that as well. So today I'm talking about emergency department visits for urgent and non-urgent ocular conditions. And I'm going to start off and give a little bit of a background about how this project came up, how I ended up doing this project. So we talked about me being a nerd. I really like research, and I really like outcomes and health services research. And a lot of times when I tell people that, they don't know what I'm talking about. So I thought it would be useful to go to Wikipedia and just review the definitions for these. So outcomes research is a branch of public health research, and it looks at the end results of the structure and processes of the healthcare system and how that impacts the health and well-being of patients and populations. Health services research is a related term. It's a little more broad. It looks at how people get access. So it looks at the system that we deliver care, how much care costs and outcomes, what happens to patients as a result of this care. So I'm really interested in this. I'm really interested in the idea of kind of what Zach and Adam talked about making what we do better. And so given my interest in this, I wanted to get more training in it so that I have more background and more experience in it. And so I learned about something called the National Clinicians Scholars Program. So it's a fellowship that I'll be doing at the end of this year. And there are four different sites. I'm going to the site at University of Michigan. And it's a two-year program. It's designed to prepare physicians for careers in outcomes and health services research. And so I'll be starting there in July. And my interest is kind of the three big projects I'll be working on during it. One of the projects is working on an electronic protocol for glaucoma management. This is actually a project that I've started working on with Dr. Crandall and Dr. Alan Morris. Another project I'm going to work on is telemedicine and its relationship with glaucoma. And then the third thing that I'm interested in is big data and how to look at outcomes for that. So this picture here is actually from the website for the National Clinicians Scholars Program. I'm hoping that they asked me to pose for this inspiring picture for the next year's website. I've been working on my, just practicing my pose for that. At Michigan I'm going to have the two main mentors I'll work with. One is Paul Lee. He's their department chair. And he's really one of the world experts in health services research for ophthalmology. And I'm really excited and fortunate to be able to work with him. My other mentor that I'll work a lot with is named Josh Stein. And he's a glaucoma specialist. And he's a health services researcher, but his real expertise is in big data. And I'm looking for a third mentor. I've asked Jim Harbaugh to be my, Dr. Mann has mentioned Jim Harbaugh. I've asked him to be my mentor working on Angry Faces. But I haven't heard back from him yet. So big data, that's another buzzword that we hear used a lot, especially over the last few years is something that's talked about a lot. And I thought it would be useful again to go to Wikipedia to really help us kind of understand. I think a lot of times we use that word and we don't really know what it means. So big data basically just means data sets that are so large or complex that the traditional means of processing data that we've used for hundreds of years, thousands of years, don't work for these. They're inadequate. And so I was really fortunate this last fall. I had an elective month and working with Dr. Petty, Chandler, Alicia, and Elaine to really schedule an elective rotation for me. So what we did was I went to University of Michigan for my elective month and worked on a research project with Dr. Stein. And so I mentioned he's a big data expert and he has a huge database of claims data that I wanted to do a project with. And so I discussed with him kind of possibilities for a project. And I decided on this topic, so use of emergency departments for ocular conditions. And this is a topic that's near and dear to the hearts of all residents. I have a picture of our call box here. Julie was using this last night, spending a lot of the night at the ER. And so I talked to Dr. Stein. This is what we decided on for my project. So a little bit of background information is that the overall use of emergency departments has really gone up. So 108 million visits in 2000 to about 130 million visits in 2010. And a recent systematic review published in 2012, they reviewed the literature and found that probably about 40% of all ED visits are for non-urgent medical conditions. So conditions that could be better handled in a different setting, not the emergency department. And another study showed that on average visits to the ED cost about four times more than comparable visits to an office. And another study, this was an interesting one, showed that non-urgent visits actually do make outcomes worse for patients that are there for emergencies. They use up resources, use the time of providers. And I think another problem with non-urgent visits to the emergency department is they're actually less convenient for the patient. And sometimes they get poorer care than they would have gotten in other situations. And so with this background, this information about ED visits increasing and non-urgent uses, talking with Dr. Stein, we wanted to really look at what does this mean for eye conditions. So what are the trends for visits to the ED for eye conditions? Are the visits urgent or non-urgent? And what factors are associated with increased utilization of the ED? And are there any things that are protective that prevent it from happening? And our real hope with this was to not just learn some kind of GWIS information, but really find things that something that could be done to make it better. So our methods, we decided to use the database that Dr. Stein has. It's the Clinformatics Datamart database. So it has records of all beneficiaries enrolled in a nationwide U.S. managed care network. And so it's, we have data from 2001 to 2014. And it's, so it's around 15 million patients total in it. And so that does qualify as big data. And for it you have ICD-9s, CPTs, age, socioeconomic information, geographic information. You have some lab information, pharmacy information. There's, it really is big data. There's so much information there. In fact, some of the regressions I'll talk about later, they take, so we have a programmer who runs them for us and they use a supercomputer and it takes the computer 36 hours to run the regression, which is pretty amazing to me to think about. So out of this, out of this database, the way we selected our patients, we chose all patients that were older than 21 years old, older than 21 years, and had at least one year of continuous enrollment in the network. And that's something to remember that all of these patients have insurance because the information comes from the insurance claims database. So no one insured patients were included. And if they didn't have continuous enrollment, we excluded them because we didn't know what was happening in the times that they were outside of the network. And then, so we looked at patients that visited the emergency department and we looked at the ICD-9 codes. So we found all of the I-related codes. And then we wanted to look at urgent versus non-urgent. So we used the ICD-9 codes and we categorized some of them as what would be definitely urgent. So things like purulent end opthamitis, hypopion, giant cell arteritis, papillodema. These were things that we said, those are pretty clearly urgent. And then we also looked at what would be really clearly non-urgent. And so we chose things like conjunctivitis, teridium, pinguecula as the primary diagnosis for an emergency department visit. And any other eye diagnosis we put in a separate category that we called other because it wouldn't be clear. And also this kind of highlights one of the issues with working with big data is that we don't have actual clinical data. We don't have the clinical nodes. What we have is what they build for. And so it's probably not perfectly accurate, but we felt like we were conservative with our classifications of urgent and non-urgent to say things probably fell in these categories. We do some regression models and I'll talk more about these later. But we looked at the predictor, so regression models to see what was associated with a visit for an urgent or non-urgent condition. We looked at age, sex, race, education level, presence of comorbid depression or dementia. And these two things were interesting things I think are important to discuss. So one was we looked at whether the enrollee has a record of frequent visits to the ED, meaning that they'd been there greater than four times a year for non-automologic conditions. So that was one group of patients. And then another was whether or not the enrollee was followed regularly by an eye care provider. So we said if they had visited an eye care provider for greater than three visits and at least six months in between, we said patients with that probably had an established relationship with that eye care provider. And then we also looked at Charles' Comorbidity Index, which is a measure of overall health. We wanted to look at some geographic variations. So we used HRRs, which are hospital referral regions. That's an accepted method of breaking up the country into distinct geographic groups. It's from the Dartmouth Atlas. And it's based around tertiary medical care. And so the United States has broken up into 306 distinct HRRs. So moving on to results, we found so out of our inclusion criteria, there were 11.1 million enrollees. And we looked at those that had visits for any eye-related complaint. And so 3.4% had a visit for this. And then I think what's really interesting is where we broke those up. So 6.7% were for diagnoses that we felt were really clearly urgent. So the most common was giant cell arthritis, orbital cellulitis, and retinal arterial occlusions. So again, we don't know if those diagnoses may have been inaccurate, but those were what the provider thought they were. And so those are pretty clearly urgent visits. We also had the non-urgent visits. So 23% of the IED visits were for non-urgent conditions. The most common of those was conjunctivitis, which was about 70%. The other 72% of visits we really couldn't classify just based on the billing codes alone. Right. Yeah. So it was the primary diagnosis. Yeah. So if they had more than one eye diagnosis, then they fell into this category. And if they had a systemic diagnosis, we didn't include those either. So if someone came in with a heart attack but also had conjunctivitis, we didn't count those, no. So this is looking at just the trends. And so this is number of IED visits per 10,000 enrollees. And so you can see the number of IED visits actually did increase through 2014, with more increase later here. And then this is looking at the proportion of visits for each year. So the black up here is the percentage of those visits that were urgent. So fairly stable. This big group in the middle is the other visits that we couldn't classify. And then the non-urgent visits actually trended down slightly over the time period. Not sure exactly what to make from that, but a slight downward trend. This was an interesting one. I bet any resident in the audience could have told you this as well. So this is looking at the daily variation of the visits. So Sunday and Saturday, by a long shot, there were more visits. Something that I thought was actually really interesting is the number of urgent visits was fairly stable over the time period. But the non-urgent visits were what really went up on Sundays and on Saturdays on the weekends. So I think Eileen experienced this last weekend and could have told us that that was true. So this is the regression model that I was talking about earlier. So this is looking at factors associated with increased risk of going for an urgent condition. So interestingly, as your age increased, you were less likely. And there's a lot of information here, so I just want to highlight a few. So as age increased, you were less likely to go for an urgent condition. The race had a little bit of an impact, not much. Ageing was a little less. And then household income, greater household income, you were less likely to go to the ED for an urgent condition. And then these were the two ones that were really interesting. So the frequent ED visitors that I described before, that was by far the greatest risk factor to say you might go to the ED for an urgent condition. And then interestingly, so having a regular eye care provider, you were more likely to go to the emergency department for an urgent condition. And so that was kind of surprising, but a couple of thoughts to explain that. One would be that these patients have a regular eye care provider. They're contacting their eye doctor and under their direction, they're going to the emergency department for an urgent condition. So they're using the emergency department appropriately. Another possibility is that these patients just have more eye problems and so are more likely to go there. And because they have more eye problems, that's also why they have a regular eye care provider. So that's for urgent conditions. Now looking at the same regression for non-urginocular conditions. And this is where I think it's really interesting, because this is potentially something we could do something about. So again, increasing age, you are less likely to use the ED for a non-urgent condition. And so as you economic, again with higher income, you are less likely to go to the ED for a non-urgent condition. And this is one that I think is really interesting. So remember, all these patients have insurance. They all have access to healthcare. So why would they do that? I think one assumption that you might jump to, and I think as a resident sometimes we jump to is, oh, they're doing it because they don't understand. They don't have the education to know. You don't need to go to the emergency department for that. But the more I thought about it, I think it's probably more related to the fact that they are in these tougher socioeconomic circumstances. They probably don't have jobs where they can just take off in the middle of the day and go, they'd be at risk of losing their job. So I think that may have more to do with it. Again, it's just an interesting pattern there. And then down here, this I think was the most interesting results we had. So again, the strongest risk factor for going for a non-urgent condition was if you were someone who had habitually used the ED in the past for other conditions. And then the regular eye provider, remember for the urgent conditions you are more likely to be going to the ED. For non-urgent conditions you are less likely to be going to the ED. Implying that having a regular eye care provider gives you access to be able to go for a non-urgent condition to have that taken care of outside of the ED. So we also looked at geographic variability. So this is urgent conditions and these are those HRRs that we talked about. They're age and race adjusted and they're per 10,000 person years. So it's not just based on population. And so I think the key here is that there's a lot of variability. Some places are using a lot that go into the ED a lot and then some places a lot less. And also that they're kind of clump geographically suggesting that there's maybe something about these areas that would be that way. This graph is called a turnip plot. And so each of these dots represents an HRR and this is to really look at the variability. So this is the number of visits. So there's some HRRs here without very many urgent and then some with more. But overall it's a pretty tight distribution for the urgent, ocular conditions. So HRR, that's a, I'm blanking on the term, a hospital referral region. So it's determined by the tertiary care center from the Dartmouth Atlas. That's kind of the publication that's in. So here again is the map of the HRRs, the 306. And this is looking at the non-urgent, ocular conditions. So again, there's really big variability and there's some areas, especially here Michigan, Northern Ohio, Western Pennsylvania where there's a lot of use. And then this is that turnip plot again. And this I think really shows the variation. So some down here, the lowest was in Medford, Oregon. And I think that was around two or three. And then the highest was in Illyrio, Ohio, which was up around 50. I looked up Illyrio, Ohio on Wikipedia. There's nothing real, that seems real special about it, the small city in Ohio close to the Great Lakes. So I'm not sure exactly what would do that. But it fits with the geographic area. There's something culturally or otherwise in that area that people are going to the ED more. Yeah, Dr. Roscoe. Dr. Roscoe, you're awesome. You're like right on the same page. So looking at these non-urgent, ocular conditions. So these are the areas that we really want to look at. Is there something about these areas? So we thought along the same lines, maybe does the number of ophthalmologists correlate with that? My hypothesis was that with more ophthalmologists out here, number of ophthalmologists per 10,000 residents, you'd have fewer ED visits because there's easier access. So maybe something looking like this. There really wasn't much correlation. We also hypothesized that maybe the number of ED physicians would impact it. So saying if there are more emergency departments, you're more likely to go to an ED. So maybe with more ED physicians there would be more visits out here. And that really wasn't a strong correlation either. We also looked at primary care providers and their role, and we didn't find a correlation there. So just a quick discussion. The study strengths, I think the biggest strength of this study is that it's a huge data set. And because of that amount of data, we were able to do these really high-quality multivariable regressions. And also we had enough data to do the geographic variability, which I think is amazing to think about all of those HRRs, that we had enough patients in each of those to be able to look at that. Another strength is that the diagnosis is given by a provider. Granted, it's not always an ophthalmologist. A lot of them are emergency room doctors. But still that's better than a number of studies in the past have been done with self-reported or things like that. Some of the weaknesses. I think the biggest weakness is that we don't really have the clinical data, that we're relying on billing data. Claims data wasn't intentional. It's not done for this analysis. It's done for billing. So I think that's definitely a limitation. But with that, I think we're able to still learn enough that we can identify some trends and some possible targets. Also, there's possibility of coding or billing error or misdiagnoses. So in summary, 23% of ED visits are likely non-urgent. So there's a significant percentage that are likely non-urgent. There's a lot of geographic variability. The ED use for eye conditions is associated with its lower socioeconomic status. The greatest risk factor is if you're a frequent ED visitor, that was by far the biggest thing associated with it. And then I think maybe the most interesting thing to me was that having a regular eye care provider actually did decrease your risk of going to the ED for a non-urgent condition. Highlighting that it's important to have an established eye care provider. Why this matters, it provides some important information. It identifies a problem for the eye conditions and then also suggests a few possible solutions. One being education, focusing on those frequent ED visitors. Possibility of doing incentives or disincentives for using the ED. And then also the idea that it is important to have a regular eye care provider. And so just in conclusion, I want to say thank you to my mentors, Dr. Stein, Paul Lee, Maria Woodward is another mentor that I'll be working with there. And then a big thanks to Dr. Petty, Alicia, Elaine, Chandler that helped make it possible for me to do my away rotation during my elective month last fall. And it was a really great experience and got some things started for me. So thank you. Any comments? Julia? So those aren't included in the ERs, no. This is emergency departments at a hospital. And we have that information in the billing. You can see what type of, how they're built. It was a curiosity question. It seemed like female sex was negatively correlated with seeking ED care. Was that true? Yeah, it was. There were even a few other things that I did there. I just highlighted a few. And I'm not sure how to explain that. Do you have thoughts? I bet you have thoughts. Yeah. I don't know how to get insurance, but they often don't want to go to their doctor. So why not go to an ER when they do have time? Right. And then, what about optometrics? Did you look at optometrics and see if optometry was correlated with higher ED visits about the serotonin provider, or was it ophthalmologist that you looked at? Yeah, so we did both. So that was combined. That would be interesting to see, to split those two up. So that's a good thought. But it was both. It was either eye care provider. Dr. Warrowska? Really cool work, right? You know what we really need to is to look at, I'm sure you've thought about this as well, but what would be the savings in terms of healthcare dollars? Right. That would be actually a big interest nation law. Right, yeah. I think for the, we're putting together the manuscript now, and I think we're hoping to do that. That's kind of a tough question to really answer, but yeah. Dr. Warrowska handling that first. I'm good. Do you have any way of knowing, and I might have blinked and missed this, but did you have any way of knowing whether these people who went to emergency rooms were actually seen by people with any eye training? So is that, is that necessary? I mean, obviously it would be very difficult to know whether the care was appropriate or the diagnosis was correct, and that would be a lot much more detailed, obviously, but how many places actually have ophthalmology trained people available in their emergency rooms? Yeah, so this did not distinguish that. So some of the diagnoses may have been provided by an ophthalmologist who came in uncalled, some may be not, and I'm not sure if from the data set we could parse that out to say who was actually seen by an ophthalmologist. Could you find out how many places that you have ophthalmologists available? Not from the data set, there may be another way to do that, and that would be, I mean, that's an interesting question. I know a few hospitals around here that don't. Well, I guess it's different between having somebody available and having somebody who actually comes in. Right, yeah. Dr. Nell, do you have something? You know, just as a suggestion, obviously you thought of this, the next step in this type of research is what do you do to intervene with the pattern, so that could be your two-year project, is how do you get official ER users to put going to the ER? You put a social worker in and say, hey, listen, there are doc in the boxes, there are practitioners that can see you, you don't need to come to the ER every time you have a minor problem. I don't know how you change that. That's the next step after the research is how do you intervene and how do you change that behavior? That's exactly what they did. They have social workers who guide that. Yeah, that's a good point. And you're right, that is the kind of the next step that's kind of different. Dr. Nell, I can do that here. Yeah. Sometimes we're looking at other health systems in the UK as you know, we repeat the point of contact, whether it's emergency or current practice or special care. There's no change of money. And so for about six months in the north of England, they did a study and found out the answer to the very question that you were raising. They introduced you to 20 pounds, which is not $32. You had to pay that. You had to pay that to whether you're in a serious condition or not. And they cut down what they call the coughs and colds, you know. You had to go to the cold waiting time. The comments missed you so. You know, it takes a lot of valuable time and resources. And they cut down the number of so-called wasted visits by about 50%. And here's the interesting thing. They did not spread this across the country because they said if you had a serious condition, like pneumonia, you couldn't afford the $25 to $25. You might not go in. So it's a double-edged sword. So the answer is that if you create an incentive not to go getting problems that you want to go with, then what happens to the serious problems? The other reason why I'm going to try to cut this out. See, new residents shouldn't complain. You don't want to be busiest, I suppose. We don't. We never complain. 11.30 to 2 in the morning. Guess why I was really busy 11.30 to 2 in the morning. And I'm not talking about a copper of coal. It's like a bit of air in the hat. 11.30, because the pubs closed yesterday. Oh, right, right. Thank you. Thank you all. Thanks, Brian.