Okay. Welcome back, everyone caffeinated and heavily sugared and all of those things. All right. We always leave that out for it there. So welcome to session 231, which immediately follows on session 225, which is the second. It's the second five eighths of the presentations by the residents and recent graduates. But before we start here, this is the fun part of my job for this here, which is to give things away. Now in the past, they come with, they would have to show up because not only would they get a certificate, but they would get their check from ACOM in the folder I'm giving, which now it's turned into kind of a quaint ritual because it's really so they get their checks electronically. So some of them didn't show up, they've just taken the money and run. But anyway, it gives us our chance to kind of recognize them as a group. And again, we're very grateful to ACOM, which has been funding this really ever since I've been doing it here for our residents to come travel and present to the AOHC. And most of them, I kind of look back on the lists of people who have presented and see all the people who are still remain movers and shakers in occupation or have become the new movers and shakers in occupational medicine. Also I'd like to present each of our presenters with their certificate, but not their check, which should have arrived electronically. You can come in, Michael. So in whatever order I got these, we'll present to first is Ashley Nadeau of the University of Minnesota for Biomonitoring of Minnesota Firefighters. I don't know, you can just wave to the crowd. Zeke isn't here with his camera, so I guess we're at a loss here. All right, Dr. Alexei Kreniev, one of the ones we still have yet to hear from the University of Cincinnati and Occupational Sharps and Needlesticks, we'll look forward to hearing from you. Another upcoming preview, Dr. Rashmi Bhuyan from the University of California at Irvine, will be presenting on Associations of Air Pollution and Hypertensive Disorders of Pregnancy. Dr. Valerie Willis, University of Texas at Tyler, again, upcoming attractions, job assignments and asbestos workplace exposure measurements for TAWP mesothelioma deaths. Thank you. Dr. Jean-Sebastien Chassé, that we heard early this morning, this afternoon, Associations between Self-Reported Burn Pit Exposures, Functional Status and Metathesis. Lawrence B.S., MD2B, for Systemic Metal Exposure and Pulmonary Function Testing. Max Blumberg, upcoming, from the University of California at San Francisco, for Cleaning and Disinfecting in the COVID-19 Pandemic, Relationship between Workload and Negative Physical Health Outcomes. Good to hear. And last and certainly not least, Alexander White, also the University of Texas at Tyler, for Automated Identification of Fibrosis Perfusion on ILL-B root chest x-rays. Thank you. One hand for all our recipients. Dr. Levine, would you take the microphone if you want? I have a really loud voice. Would you take an opportunity to express our gratitude to you, Dr. Meyer, for doing this year after year and raising the bar for disability for occupational medicine residency training for our clinical participants? Thank you, except I'll disagree with the one point. It's the residents and the mentors who are raising the bar for it and not me. So remember, just as we start presentations, that the mic should be up front there. And we have one last announcement. I wanted to welcome Dr. Pamela Kral, who is representing here the ACOM. She wears many hats, but is representing the ACOM History and Archives section, who had a brief announcement and wanted to recognize their student essay winners. Yeah, we were really excited this year. It was the first time we've done an essay contest through the History and Archives section open to all students. And we want to thank ACOM for providing the support for the prizes for them as well. And we needed a venue to be able to recognize folks up here. So we had two categories, a virtual category with a virtual registration for the conference and an in-person category. So our winners for the in-person category from University of Maryland, Magdy of... I can't. Help me with your name. Thank you. Thank you so much. And you can see his title, James Jim Moran, Hero of the Philadelphia Labor Movement. Congratulations and nice work. Look for their essays. We're going to try and get them out so that people can... I learned a lot reading them, and I'm sure everybody would enjoy seeing them. So thanks. And next, we have Justina Su from Ohio University. Justina, please come on up. Nice, nice job. Also selected Jim Moran as well in terms of the topic. So the legacy of Jim Moran. Again, be looking for these. So we'll find out who Jim Moran is. Exactly, exactly. So thanks so much. Nice work, and we'll see you at the History and Archives section meeting. Great. And then last in the virtual category, we had Karina Kaufman from Uniformed Services University who wrote about Lewis Hine, and thank you for the Lewis Hine photo. Such a history of labor geek, but most of you probably recognize many of Lewis Hine's pictures there, and that's one of his children as a cotton spinner, and a lot of very relevant industrial photographs. Essentially, he was Earl Dodder's predecessor, and they kind of continue on the same. But go ahead. Any last words? No, I just say congratulations to everybody. We hope to grow this in future years, so please spread the word. We'll find a different geographically themed topic that has to do with Orlando next year. Thanks. Thank you, Dr. Kral. Greatly appreciated. Congratulations to the winners of that. I think that's an excellent group to be part of there. So who is first up here? Dr. Max Blumberg, the University of California at San Francisco, is going to talk about something very relevant in my group here, but that we should be thinking of all completely. So what happens to cleaners in general and what happened to them during the COVID-19 pandemic? So welcome, Dr. Blumberg. Thank you, and thank you, everyone, for having me. So this is the title of my presentation, Cleaning During the Pandemic and the Relationship Between Workload and Negative Health Outcomes. I don't have any personal financial disclosures. I did want to thank the NIOSH grant that helped support this research, and this research was done through the Northern California Center of Occupational and Environmental Health, the ERC that's between Berkeley and UCSF. We also had collaborators in the California Commission on Health and Safety and Workers' Compensation, the Service Employees International Union, and a nonprofit watchdog organization called the Maintenance Cooperation Trust Fund, and the project was approved by the IRB at UCSF. So first, a bit of background. There are approximately 2 million janitors in the United States, with about 200,000 of those in California, and janitors across the board have high physical demands and potential hazardous exposures at work. Those primarily include musculoskeletal injuries, but also potential for respiratory and skin exposures, and as a profession, janitors are ranked third in the work-related injury and illness rate. There's about 35,000 cases reported each year of work-related injuries or illnesses, and that's likely a gross undercount due to the fear of retaliation at work. There's no prior studies on mental health of janitors, so I'll talk a bit about that. This was a novel aspect of this project, and the impetus for this project was related to the pandemic causing increased disinfection protocols at work for these workers that were already at a high physical demand before the pandemic, and my specific project is part of a multi-component study, which I'll talk a bit about later. So the objectives of my project were to describe the workload of janitors in California and the negative health outcomes that they face, and it was collected via a survey done in 2022, assess the association between these two things, and develop recommendations to improve this work environment. I'll go through this part by part. This is the methods of the project. So as I said, it was a survey-based study, cross-sectional analysis. The survey was sent out to 40,000 janitors in workplaces across the state of California and working in many different environments. The survey was sent in both Spanish and English, and it was sent via email and text message, although there was also an option for janitors to fill it out over the phone if they would like. Over 1,000 people responded to the survey to some extent, although the survey took about 20 minutes to fill out in total, so not everybody answered every question, and for the questions that were relevant to this analysis, the N is about 400. And the questions included 16 common tasks that janitors do, things like vacuuming, dusting, et cetera, and there are different ways to characterize the workload that janitors face. So typical intensity and peak intensity, the second and third underlying terms here, are common ways that the literature of ergonomics looks at workload in different types of jobs, and the workload index is a less commonly used method of characterizing workload, but we felt that it was important because the intensity is only a part of the picture. So workload index is a term that refers to a multiplication of the intensity, frequency, and duration of a given task. So, for example, how would you rate your workload when you're vacuuming on a 0 to 10 scale? How intense is it? How often do you do that task? So, you know, every day, once a week, once a month, and when you do that task, are you doing it for one hour at a time or five hours at a time? And then you multiply those three together. It gives you an arbitrary unit, but then you can use that to tertile the exposures and make a low, medium, and high exposure group. Typical intensity refers to the intensity question alone, and it's the intensity of the task that's done for the most time. So it also uses the frequency and duration, and whatever task they do the majority of their time, that intensity is rated as the typical intensity, and again, that was tertiled. We wanted to tertile the peak intensity as well, but as you'll see in a moment, most janitors' peak intensities are very high, so the distribution didn't allow for tertiling, and we just did a median split for that type. The outcomes are a little bit more straightforward to understand, so we looked at five different outcomes. The first is pain measured in four body regions that you can see on the top of the slide there. The janitors were asked to rate their pain on a score from 0 to 10 in these four regions, and then those scores were averaged, and if they had a score of 5 or higher, they were considered to have severe pain. Their pain medication use was considered regular if they used either over-the-counter or prescription pain medications at least one week per month. They were considered to regularly miss work because of that pain if it happened at least once every other month, and we made work-related injuries a binary category, either at least one in the last year or none. And then we used previously validated questionnaires for anxiety and depression, the GAD-7 for anxiety and the PHQ-9 for depression. So we took all of those exposures and outcomes, used logistic regressions to generate odds ratios and 95% confidence intervals, and we assessed reconfounding by age, sex, smoking, education, and comorbidities, but only age and sex changed the odds ratios by at least 10%, so we did not include those other variables in the final models. Moving into the results, you can see here that the vast majority of the individuals were Hispanic, 95%. The age distribution is roughly 50-50 in the groups 30 to 50 years old and 50 to 65, with few members over 65 or less than 30, and about three-quarters female. So these, again, are the three different ways you can characterize workload. As I mentioned, the mean intensity that the janitors are reporting is extremely high, 8.3 out of 10, and that's why we were unable to tertile that specific way of looking at workload. Typical intensity, again, is also quite high, and then workload index, as I mentioned, has an arbitrary unit and it's only used to do the tertiling. This is a summary of the outcomes. So about 50%, a little more than 50% of the individuals reported having severe pain. A little more than 50% are regularly using pain medications. About 20% are missing work regularly because of that pain. About a third have at least one work-related injury, and about 17% have either anxiety or depression. And then the next several slides are all in the same structure. It's a little busy, so I'll walk you through it. The top left corner is the identification of the association that's being looked at in the model. So this whole slide refers to severe pain. As you go down, there are the three different types of ways of categorizing workload, and you can see the tertiles and the median split for the bottom third. As you go across from left to right, the capital N refers to the total number of individuals that are in that tertile. So, for example, the 117 refers to the fact that there were that many individuals who filled out the questions relevant to this logistic regression, and the lowercase N refers to the people who are considered cases. So 39 of those 117 reported severe pain. And then for each means of characterization, the low exposure group is used as the referent, and the median and high workloads are compared to the low. As you can see, there's pretty extreme odds ratios on this slide, not something that you see in the epidemiology every day, although the confidence intervals are quite wide due to the low ends. And they're all bolded because they're all statistically significant. And finally, I'll just say that the odds ratios increase as you go down the slide, so that suggests a dose-response relationship. There's basically the same story for all of these. There are a few categories that are not statistically significant, but in general, severe pain, medication use, missing work, having work-related injuries, and having anxiety and depression are all in some form of the categorization statistically significant in a dose-dependent fashion. So to summarize everything I just said, consistent with previous literature, we found a high burden of workload. And for the physical health outcomes, consistent with previous literature, we found a high burden of negative health outcomes. And new to the literature, we found a high burden of mental health conditions as well. And for each of the different ways of categorizing exposure, the odds ratios show a dose-dependent response. So the next steps for this project are to complete a sub-analysis of everything that we've done, but breaking it down into the 16 tasks in order to hopefully identify some of the tasks that are the biggest culprits. And as I mentioned, this is a multi-component study, so there are collaborators of mine who are looking at different aspects of the survey that talk about individuals who might work a second job where they're working up to 80 hours a week, whether they're caring for someone at home that requires additional physical labor, looking at effect modification by age or sex, and a different set of variables that could add to the potential recommendations to improve the environment. And that is all. Great. Thank you very much, Dr. Blumberg. Some questions? Dr. Klor? Hi. Nice job. I thought it was a very good presentation also of your data. I have a question. Can you clarify who assigned the intensity? Was that self-reported intensity? I think that might be a limitation. Yes. So that is definitely a limitation. It is based off of self-reports. There is a follow-up or a second component of the study that is actively recruiting right now that is using wearable devices and video monitoring to objectively measure the workload as well. Great. Thank you. Hello. Very nice presentation indeed. My question is actually related to the previous question. So how did you classify the peak intensity of pain and peak intensity of anxiety and depression? So the peak intensity just refers to the exposure. It doesn't refer to the intensity of their anxiety or depression. The GAD-7 and PHQ-9s are questionnaires for measuring or screening for anxiety and depression, and they have pre-established cutoffs based off of those questionnaires for whether they qualify. It becomes a binary outcome. Just one last question since it's obviously you're in a state university and much of the funding comes from that and, of course, with yours there. Do you have a venue through which you might be able to help? And you're also working with the unions there. Do you have a venue through which these results will be disseminated, reported, acted upon? Yeah, so the union that I mentioned, SEIU, Service Employees International, represents about 40,000 of the 200,000 janitors in California. So they helped us to distribute the survey. and then there were additional non-unionized members, about a quarter of the survey respondents came from other non-unionized workplaces, but we will be disseminating the results via the union. Yeah. It sounds like it needs action. Yes. Absolutely. Yeah. Thank you. Okay. Turn it up next here, and get that going. So we'll next welcome a little bit of change of intellectual venue here. We have two excellent presentations on asbestos in the area around Tyler and in Texas here. We'll first welcome Dr. Valerie Willis from the University of Texas at Tyler, who will be speaking on job assignments, workplace measurements for mesothelioma deaths. So welcome, Dr. Willis. Thank you. So again, I'm Valerie Willis, and I'll be presenting this presentation here. And so the funding for this presentation came from the Jesse Jones Distinguished Professorship of Occupational Health Services Endowment, held by Dr. Levine, that kind of helped fund some of the collection, the administration data collection assistance for this project. And then, of course, just the residency program itself, the grant that helps fund our residency and pays my salary, which is great. And then we had IRB approval from the UT Health East Texas IRB office. And I have no disclosures. I make no money off of any of this other than my salary for residency. So the purpose, asbestos workers have a higher risk of developing mesothelioma. However, few studies have looked at specific jobs and job locations within asbestos factories. The purpose of this study was to investigate the asbestos exposure in different job locations of the Tyler asbestos plant to determine if there was a relationship between the durations of exposure and the air fiber concentration while working in those locations among workers who developed pleural versus peritoneal mesothelioma. So brief introduction. So Tyler, Texas had an asbestos plant that operated from 1954 to 1972. And during that time, 1,130 workers were exposed to amici asbestos while creating pipe insulation. These workers worked in one of six locations throughout the plant and had exposure to asbestos that we later found with OSHA, was over the OSHA standards that were established in 1970. But according to the standards, we had some air surveys that were performed in 1967, 1970, 1971, that when we looked at them, they were definitely over the later OSHA standards. The workers were enrolled in a health surveillance program to monitor for disease development as a result of their exposure. And previous studies identified 23 workers who developed mesothelioma from the 1,130 workers that were enrolled in the TAWP health surveillance program. So this was a cross-sectional study using the TAWP database to compile secondary data on the 23 workers who died from the mesothelioma through 2011. That's how far we had death certificates for. The job titles of the 23 mesothelioma cases were organized into one of the six location categories depending on the job title and organized based on kind of the operation categories as described in an article by Johnson et al. in 1982 and the air survey data reports. So you can see most of the jobs that these 23 workers had kind of end up falling within the forming, finishing, and the miscellaneous categories. The time the workers spent at each location were calculated and then estimated. From that, we estimated the fiber burden for each time, for each job location using the air survey means from 1967, 1970, and 1971. And we had those mean fiber calculations or the mean fibers from tables drawn from articles by Hearst et al. and Johnson et al. And we assumed that there were no other external asbestos exposure. And finally, we calculated the total fiber burden for each worker and for each job location. So methods, we kind of did some preliminary analysis using just Microsoft Excel and then statistical analysis, we used IBM SPSS statistics with an alpha of 05 to kind of compare the days worked and the estimated fiber burden of the pleural mesothelioma cases versus the peritoneal mesothelioma cases. Then we also decided to compare the days worked between the six jobs and the estimated fiber burden between the six job locations. So in going into the results, this is kind of a table just displaying the details for the days worked by each worker in each of the six locations throughout the plant during their time employed. We could not obtain the job data for two of the workers who developed the pleural mesothelioma, which is case five, case 22, but we did have their employment dates and could calculate kind of the number of days they were exposed to the asbestos fibers. Table four similarly contains the estimated fiber burden for each worker for each location in which they worked while they were employed at the asbestos plant. The graphs on this slide and the next kind of provide a quick look kind of at the totals of our preliminary data. This graph one shows the total number of days worked in each job location and the number of workers who had worked in those locations, kind of the total number. And the job locations were arranged in order of least to greatest, so you can kind of see the progression of kind of the pattern there. And so this was the days worked. And here's the estimated fiber burden, the overall total estimated fiber burden for each job location and the number of workers who had worked in those locations. The jobs were again arranged kind of in order of least to greatest total estimated fiber burden. And again in this kind of you see between the two slides kind of the forming and the miscellaneous sections kind of both were high in both sides, so kind of quick back and forth and then to the next. To analyze our data we used two sample t-tests to compare the days worked by workers who developed the pleural mesothelioma to those who developed the peritoneal mesothelioma. This table kind of shows our comparison for the total days worked and we found that there were no significant difference between the two populations for any of the locations. We also used the two sample t-tests to compare the estimated fiber burdens of the workers and this table kind of shows that comparison again and again there were no significant differences between the two. Finally, we also decided to do kind of a comparison between the different jobs, the six locations. So we used one way ANOVA test and then kind of did some pairwise comparisons to compare those six locations and the total number of days worked and the total estimated fiber burden in those locations. So when we compared them we found that there were no significant differences even when we decided to divide up them by the pleural versus the peritoneal kind of separately and kind of compare the job locations for just the pleural cases and then just the peritoneal cases and still there was no significant difference at least in terms of the total number of days they worked. In comparison we looked at the total estimated fiber burden in those six locations just comparing the total cohort there were no significant differences but then when we kind of divided them up again with the pleural cases there still was no significant difference but then with the peritoneal cases we did see that there was there seemed to be a difference. So the difference may exist in that forming location when we actually kind of parsed it out that forming location where we had seen earlier in the graphs where there was a high amount of people working in those locations and also at high exposure. So there was probably something different just about that forming location where possibly those patients were maybe more likely to possibly develop peritoneal cancer or peritoneal mesothelioma. So kind of in a random thing that I just decided I wanted to kind of map out how the different workers and their days kind of so I decided to plot the total number of days that each worker had worked versus their total estimated fiber burden for each of those workers and kind of had each of their colors or had them divided up kind of stratified by the pleural versus the peritoneal mesotheliomas that they developed and kind of just graphed it to kind of see if there was possibly a line of fit that we could maybe match to these patients and then since we were missing the two patients' job, two of the workers didn't have their job titles we were wondering if we could maybe use these lines of fit to maybe estimate how much exposure they could have had using these fit lines. So most of the workers who developed the mesothelioma, 73.9% spent some time working in that forming area of the plant. The forming area also had one of the highest asbestos exposures in the three air quality surveys as we had seen in table two and the area, this area was identified to have the highest overall estimated fiber burden of all the locations. The miscellaneous area had the highest number of days worked of all the locations but only had a few workers actually work in that area so only 21.7% actually spent time in that location that we are aware of. Of the 16 workers who developed the pleural mesothelioma, most of them spent that time in the packing and miscellaneous areas and most of that, their estimated fiber burden came from those same areas. For the seven workers who developed the peritoneal mesothelioma, most of them spent time working in that forming area and the miscellaneous locations as well and again, most of those, their estimated fiber burden came from those same locations. However, the forming location had significantly more asbestos exposure likely resulting in a much higher estimated fiber burden than the other areas of the asbestos plant. So again, we found kind of no significant differences when we were comparing just the pleural versus the peritoneal cases and there are no significant differences between the six job locations except for when we parsed out just the peritoneal cases and they had that, maybe that forming location was probably a little bit more indicative of developing mesothelioma. Our data had some limitations, the most obvious being that our sample size was very small. We only had so many cases to work with and we were kind of limited to the data that we had that existed in the TAWP database that had been collected up to 2011 and we had been missing the two cases, the job titles for them. We also assumed that there were no, there was no other asbestos exposure for outside of the Tyler asbestos plant. We didn't know if they had any other outside jobs that might have exposed them or environmental exposure to asbestos. So we just, that could be a limitation in our data. And finally, when we had been considering this project, we were thinking about maybe looking at PFTs and so we were maybe concerned about their smoking status and their history, but then we found out the history just was very incomplete and we were unable to use that data. So probably could have been useful in the future. In conclusion, to our knowledge, our study is the first kind of to examine the jobs and operation locations within the Tyler asbestos plant and assess the asbestos exposure in those locations for workers who develop mesothelioma. The work location and the time spent in the location together seem to be a higher predictors of the increased likelihood to develop mesothelioma rather than just the exposure location and time spent in a location separately. There was no significant difference between those who developed the pleural versus peritoneal mesothelioma in our small sample size. Additionally, the forming location may have contributed more to the development of the peritoneal mesothelioma than for the pleural mesotheliomas. The results from the study just kind of reiterates the strong association between kind of occupational asbestos exposure and mesothelioma enhanced by both the concentration of respirable levels as well as duration of exposure. Did I go backwards on that? Yeah. Finally, we concluded that there was no apparent dose response relationship for either the mesothelioma types since kind of both occurred at lower exposure burdens. And all the references and any questions? Thank you. That's good. All right. Dr. Hartz? Thank you for your talk. I'm from Southeast Texas. I didn't even know there was this asbestos plant in Tyler. Yeah. It closed in 1972, so long gone. So, as you kind of got into this field and doing kind of your background research, kind of what are the distinguishing risk factors that we're seeing for peritoneal versus pleural mesothelioma? Like, all I know to tell a patient is asbestos. I mean, but are there, as you dealt through this, like what things might predispose to being peritoneal in kind of disease manifestation? Well, really, if we had been able to have the smoking data, it was actually the smoking data that we were hoping that we could maybe have used to kind of show maybe association. But it seemed like a lot of the workers, at least from this plant, did at least smoke at some point in their career while they were with the plant. But we couldn't tell how long they had smoked or when they quit, if they had continued smoking again. So, we just couldn't really use that data. So, unfortunately. I didn't know smoking was a risk factor for mesothelioma. Not for mesothelioma, but for asbestos, yes. For health effects. Yeah, for health effects, because for the respiratory cilia to remove the asbestos, it just doesn't work so well when you're smoking as well. It just kind of kills all those cells, and then the asbestos stays there. And that's why they're more prone to develop mesothelioma. Thank you. Great. Thank you. Other questions? A quick question, I guess, for that. Do you think, given that your graph also sort of bears out, though not enough time to say it, that you've got probably some low-dose brief exposures that are congruent with what we know about mesothelioma anyway, which is there can be sort of short-term, maybe short-term high-intensity, but really short, not lung cancer, not asbestosis. But that clearly bears that out. Do you think there's any value in trying to quantitate this any more like this? In other words, are numbers going to help, or is it just you've got the one particular area, and that looks worse than any? I mean, numbers are appalling here, because they're obviously sort of even post-OSHA. And so all you can do is sort of look at it in retrospect. Do you think it's just worth looking at that more qualitatively and thinking about these people who are at risk because they were in that high-intensity area that might be more worth than trying to quantitate fiber burdens? Yeah. It'd probably be more useful to look at the different locations again, kind of since we kind of see that there was that forming area that possibly had just a higher exposure, maybe looking at the outcomes of the other thousands of patients that we had in our study. We definitely didn't have the time or the money to do that for our study. Not for a resident. Not for a resident within the two years that you're at the residency. So that might be something that we might be able to do to kind of look and see what kind of outcomes they had in the forming location if they had worked in the forming area. So you've got your own dead horse to flog for a while. Yes. Or future residents. Future residents. Dead horses. Right. I can see the gleam in Jeff's eye. Thank you. Thank you. All right. Let's get these up here. So, Dr. Hines, just to emphasize there was asbestos in eastern Tejas. We'll have here Dr. Alexander White, who's going to talk to us about machine learning and automated identification of fibrosis and perfusion. Dr. White is our second resident coming to us from UT Tyler, which again puts the lie to Jeff's statement that it's me who's like upping the game in this session and that it's really the residents and the program directors. So welcome, Dr. White. Thank you. Thank you very much. Again, I'm Alexander White. Happy to be here to present this. I'd like to thank Dr. Rowlett, my mentor for this project, who helped me navigate the Department of Energy R&B process, and we gained approval from them. They also provided funding for part of the program and part of Dr. Rowlett's salary, I believe. I don't know that for a fact. You're not required to read that. Just in general, we have other grant funding that provides for my salary, although I was not really personally enriched by this research. So worldwide, there are 60,000 new cases of pneumoconiosis every year. Since 1950, the ILO has periodically published guidelines on how to classify chest X-rays for pneumoconiosis as part of their efforts to combat these diseases. The reports are colloquially called B-reads, and on the right-hand side is an example of a B-read form. One subsection of the B-read is an assessment of small lung parenchymal opacities, which are of particular concern for pneumoconiosis surveillance. These are graded with a perfusion score, which is a measure of the concentration of these small parenchymal opacities. Our studies have shown poor intra- and even inter-relater reliability when it comes to grading perfusion scores. There is a national shortage of B-readers, and it would be very nice if there were an automated solution to help alleviate this shortage. For this research, I used machine learning or artificial intelligence to train a deep convoluted neural network to automate the identification of abnormal perfusion. The chest X-ray test set for this study was derived from medical surveillance records of 1,300 patients from the Department of Energy's Pantex former worker program. The Pantex data set was not quite large enough to effectively train an entire neural network on its own. So, for that purpose, I used the NIH chest X-ray 14 data set, which consists of over 100,000 images from over 30,000 patients. Importantly, the NIH data set includes a pulmonary fibrosis label, amongst other labels, with about 1,000 positive fibrosis studies within their data set. This number was sufficient to allow me to train the neural network for my purposes. In broad strokes, training and validation of the neural network was done using the NIH data set. And after training, I went ahead and I used the model that had been made off that data set to interpret the Pantex test data set images. And we compared the results of the official ILO B-reads to the predictions made by the neural network. We use a process called transfer learning, where we start with a neural network that has already previously been trained to interpret images based off the ImageNet data set of over 14 million images of commonplace items, like dogs and cats. We keep parts of the neural networks that are good at identifying small features from these images, these features like edges and small shapes, and we retrain the final layers of the neural network to classify images into our categories of interest, namely abnormal versus abnormal chest X-rays. The model based on ImageNet data set have previously been used for identification of cancer on pathology slides, and also for finding COVID pneumonia on chest X-rays. I pre-processed the NIH data set to ensure quality, including limiting age, excluding follow-up normal chest X-rays, and removing obvious misclassified images, such as lateral views that were erroneously included in the NIH data, which should have been only of frontal images. To account for the training set being unbalanced, with many more normal chest X-rays than there were of fibrosis, I used a combination of techniques called class weighting and oversampling. For the pre-trained neural network, I chose the Inception V3 model because of its computationally – its combination of low inference computational time and high native resolution and high test accuracy on the ImageNet data set. Really, most importantly, it was the model that I could use that would run fast enough on the cloud so that I could run my model overnight and not over, like, four or five days. The new classification had the following architecture. First, there was a pooling layer, followed by a dropout layer, then a fully connected layer, and finally a softmax layer, which, at the end of all that, yielded a probability estimate for whether the image being interpreted was normal or abnormal. The results of the model fitting yielded a best validation accuracy of 76 percent at epoch 19 and validation loss of 0.5 at epoch 14. The best accuracy and loss of the training set was better at 79 percent and 0.45 percent respectively at the 50th and final epoch, but the validation metric did not improve past epoch 20. The neural network weights at epoch 19 were selected because of the high validation accuracy. The initial application of this model to the Pantex test set yielded a prediction accuracy of 79 percent. The COA-CAPA score was 0.6, which indicates a model that is fair to moderately reliable when compared to the official b-read. However, the area under the curve of the ROC curve, or the receiver operator characteristic curve, was quite high, suggesting that the model could yield even better results with a more appropriate threshold value. These plots show the neural network's prediction probability score for each tested — each of the Pantex image tests, and whether the results were — that it yielded turned out to be a true or false, negative or positive. When using a default threshold of 0.5, the model predicted a string of false positives results, and there was a large gap between the first of these results and the default threshold. So, when the threshold value was adjusted to 0.156 from 0.5, that minimized the number of false negatives to positives that I was getting, and the model accuracy improved to 91 percent with an associated COA-CAPA score of 0.8, which is an excellent reliable — sorry, an excellent result. In conclusion, we used a large public data set to train a neural network to predict abnormal perfusion on chest X-rays. The ILO B-READS that we used as the test set are much more standardized than a regular radiologist interpretation. These B-READS are an underutilized resource, in my opinion, in the medical imaging analysis field where obtaining a good ground-truth label is a significant hurdle. If a big data set of ILO B-READS could be collected and released publicly, like what NIH has done with their data set, I think that would be a real boon for researchers and it would be a great asset for the medical and, honestly, the computer science community. I have listed some of the limitations of this research here. One was that the small size of the Pantex data set. This necessitated using the separate NIH data set for the training and testing and limited the predictions that this network could make to just a binary assessment. If I had a sufficiently large data set to work with, I believe that I could have made like a full perfusion score prediction instead of just an abnormal versus abnormal assessment and maybe even looked at other things like abnormalities outside of the lung parenchyma such as pleural abnormalities. So that was my research. Any questions? Thank you. Dr. White, as I mentioned, it made me feel like an old dog trying to learn a new trick here, but it looks like we've got younger members of our audience asking the questions now. Yeah, hey, I'm Ross Mullinax, Naval Medical Center, Portsmouth. I oversee a large naval shipyard where we have a pretty good cohort of asbestos workers. So this, although I don't work at the shipyard primarily, this question does come up, right? I had a conversation recently with one of the docs there who was basically complaining about the B-reads. We'd had a number of kind of young workers new to the program that got their B-read and it was a wildly over-called and they subsequently had a CT scan that was normal, and so he was kind of complaining about the value of B-reads. So I think it's interesting, your potential solution of going to something more reliable based on kind of AI type stuff, but the question also, I think, and this is a little tangential to your research, but what about the possibility of, I think B-reads were, I know I'm going a bit long, sorry, I'll get to the point, B-reads, they created those a long time ago, and now with the wide availability and the resolution of CT scanning, should we just kind of punt on B-reads altogether and use something? They certainly have their limitations. I was fortunate enough to be involved in a DOE discussion a couple of months ago, also kind of on this topic, B-reads aren't done all over the world, they have different systems in different countries for interpreting chest X-rays, which are simpler than B-reads, B-reads are pretty complicated and perfusion scores are really difficult to do. So I think there's still going to be a role for chest X-rays, maybe as a screening metric. I know CTs are coming down in cost, it's almost, there was a presentation earlier. And exposure too. Yeah, and exposure is also coming down quite a bit. So there was a presentation on the first day of the conference advocating using CTs instead, but CTs aren't going to be available all throughout the world, throughout Asia, Africa, we're still going to be relying on chest X-rays for a long time, I suspect. But I do think that artificial intelligence is going to kind of play a role. Next. Hey, Alexander, Ashik Zaman, ExxonMobil. First of all, a recent graduate of the program in Tyler, and I'm just so impressed with the research that you guys did, and this is kind of a related question, this is a pretty advanced topic and it comes up in the corporate world a lot, how we apply artificial intelligence. I'm just wondering how you got into this, how did you pick this as your project? Thank you. So my undergrad training was in computational physics, which is physics with a lot of computer science. So I had some background in it already. I hadn't done anything like this in probably a decade, and then I started looking back, and it's really amazing what's happened in the last decade in this field. The libraries that I use are all publicly available, they're published by Google, and it's so much more accessible than what it was when I was training. And what they're doing right now with like ChatGTP, I know there's professors in the audience here, if you aren't ChatGPTing proof your courses right now, you're going to be in for a shock. So AI is coming, and it's going to be very interesting. Last question quick. Yeah, sure. It'll be quick. I was just wondering, given more time and resources, could you improve the power of this system to be more accurate? With more resources? I mean, the number one resource that I could ask for would be a bigger data set. DOE has all these images, but they're all siloed. So I just had access to the Pantex images, but there's former worker programs all over the country and other DOE depositories of images. And if the DOE, and just in general, if we could lump all the images together, make them publicly available in a way that they've been anonymized, that would be a really great thing for people to work on. Cool. Sounds like a cool next project. By the way, Dr. White, we're taking your advice, and we're heading back to pencil and paper exams in our courses now. Nobody has to worry about ChatGPT any longer, at least. All right. So coming to us from the University of California at Irvine, Dr. Rashmi Bhuiyan, who's the resident there and is going to discuss associations of air pollution, hypertensive disorders of pregnancy. So welcome. Thank you so much. Hello, everyone. I'm Dr. Rashmi Bhuiyan. I'm a resident in the University of California, Irvine, and it's a pleasure to be here. And thank you, everyone, for being here. This is a part of my thesis project. This topic is dear to my heart as a practicing OBGYN from India. Also, I would like to express my gratitude to Dr. Wu, who is the director of the graduate program in the university, and a postdoc, Sun Yi. They helped me a lot in collecting the air pollution data, which is relatively new to me. This is a little bit about the financial relationship disclosure. This study was supported by the National Institute of Environmental Health Sciences, and we obtained both the IRB approval and the UC institutional approval, ethical committee approval to conduct this study. I'm sorry. Objectives of this study was to investigate the relationship between HDP, hypertensive disorders in pregnancy, in short, HDP, and maternal residential exposures to PM 2.5 and its five constituents, and it is based on a large population-based pregnancy cohort on the Southern California Kaiser Permanente patient population. We obtained the data from using the Kaiser electronic health record system for the period of 2008 to 2017. Another objective of this project was to compare the effects of PM 2.5 and its constituents on gestational hypertension and preeclampsia, eclampsia separately. So to give you a little bit overview, the HDP, hypertensive disorders in pregnancy, it is classified, probably you know about it already, by ACOG, American College of Obstetrics and Gynecology, into three broad categories based on the severity. Gestational hypertension includes the cases with isolated elevation of blood pressure above 140 over 90 beyond 20 weeks of pregnancy, and it is not associated with any endorgen damage. And whereas the preeclampsia is a condition where there is elevation of blood pressure after 20 weeks of pregnancy, and it is associated with either proteinuria or low platelet count or liver enzyme abnormalities or renal impairment indicated by raising creatinine level. And the preeclampsia is the most severe form of HDP when the preeclampsia is associated with the neurological features and seizure is the hallmark for eclampsia. So for the purpose of this study, I divided the cases of HDP into milder category, comprising of the cases only with gestational hypertension, and the severe category comprising of cases with preeclampsia and eclampsia. And the hypothesis of this study was that increased level of PM2.5 and its constituents are associated with increased risk of developing HDP, and that the associations between the HDP and air pollution will differ by milder category of HDP, that is, gestational hypertension, and the severe category of HDP, that is, the cases of preeclampsia and eclampsia. It is a retrospective cohort study based on Southern California Kaiser Permanent Hospital electronic health record system, and it includes 15 different hospitals in Southern California. And participants were single-term pregnant women giving birth from January 2008 to December 2017. And the PM2.5 total mass and its five constituents, those are sulfate, nitrate, ammonium, organic matter, and black carbon, monthly concentrations in ambient air was measured for the period of 2007 to 2017, and it was obtained by using fine-resolution geoscience-derived models developed by Dalhousie University, Canada. And this is the flowchart of my research design. Initially, we caught 448,846 pregnancies. Then we excluded some of the participants based on the exclusion criteria, like those participants who are not KPSC members were excluded from the study, and those with gestational age less than 20 weeks or more than 47 weeks were excluded. And those who did not have address data or air pollution data, they were excluded. And also, multiple pregnancy cases were excluded from this study. We considered only single-term pregnancies for this project. And we ended up having 386,425 pregnancies for the severe preeclampsia-eclampsia cohort, and 373,969 pregnancies for the gestational hypertension cohort. We got a little bit less in the gestational hypertension cohort because the cases with chronic hypertension were excluded from this cohort by definition. Then descriptive analysis was conducted. I did not include the table because it was a very long table, could not fit it here, but I have included the main results from the descriptive analysis. The prevalence of gestational hypertension was seen to be 4.8 percent, data of preeclampsia-eclampsia was 5 percent, and both of these conditions were seen to be highest among Hispanic mothers for gestational hypertension, 44.1 percent, preeclampsia-eclampsia, 54 percent, followed by non-Hispanic white mothers. And we did not see noticeable differences in frequency of any of these conditions based on maternal education, household income, passive smoking status, and primary priority. We did not see any noticeable differences based on maternal pre-pregnancy low BMI, undernutrition mom, normal BMI, or overweight BMI, but frequency of both these conditions were increased in obese moms in all three categories of obesity, obese class one, class two, and class three. This is the main analysis of my study. Multivariate Cox regression model was conducted separately for the milder category and the severe category. Here in the milder as a gestational hypertension cohort, we see that there is negative association to develop the milder disease in exposure to all these PM2.5 constituents, but these values are not statistically significant because it included confidence interval included one. And on the other hand, coming to the pre-eclampsia-eclampsia cohort for the severe disease, we see the positive association on exposure to all the PM2.5 constituents, and the positive associations to exposure to PM2.5 total mass, organic matter, and black carbon were seen to be statistically significant. So we started the risk of pre-eclampsia-eclampsia are positively associated with PM2.5 total mass and to other PM2.5 constituents, and the risk was highest in exposure to black carbon, hazard ratios 1.11, followed by organic matter, followed by, I'm sorry, PM2.5 total mass, hazard ratio 1.07, and then organic matter, hazard ratio 1.05. Then subgroup analysis was conducted based on the race, pre-pregnancy maternal BMI, smoking, and season. For the gestational hypertension cohort, we saw that Asian moms with normal pre-pregnancy BMI were at least risk to develop the milder disease. And the mothers from household income, more than 71,591, was at least risk to develop the milder disease in exposure to black carbon. Then coming to the smoking subgroups, interestingly, current smoking moms are seen to be at least risk to develop the milder disease, followed by past smokers and never smoker moms in exposure to black carbon. It is an interesting finding. It is a paradoxical effect of smoking on hypertensive disorders in pregnancy, and this fact actually has been established by other current literatures. And then coming to the severe subgroup cohort preeclampsia, we saw that mothers from household income less than 43,667 are at highest risk to develop the severe HDP. And we did not see any significant differences to develop the severe disease based on the race or ethnicity, smoking, and maternal pre-pregnancy BMI. In the conclusion, this study is the first study to examine the associations between PM 2.5 constituents and the gestational hypertension and preeclampsia and eclampsia separately, and it was a large and diverse population-based study. The other literature that exists on the association, they considered the HDP as a spectrum diagnosis. They did not classify this as the milder and the severe category. As a result, most of the other studies' results are inconsistent. Some of them show the null association, or some of them have even showed that they are protective to HDP. But if you classify milder and severe, this is not a fact. We saw that the exposure to PM 2.5 total mass, organic matter, and black carbon are associated with increased risk to develop the severe disease preeclampsia and eclampsia. Some of the weaknesses of this study is that we considered the outdoor pollutant exposure based on the maternal residential zip codes, and we did not consider indoor personnel level exposures. Also, data was collected using the electronic health record system, so there are some missing data, and there is some risk of inter-examiner variability in data collection. And for the future reaction of this study, more studies should be conducted separately, of course, on the milder and the severe categories of HDP. And some socio-demographic data can be collected using surveys. Instead of a health record, it will give more accurate information on the socio-demographic profile. Also, indoor pollution exposure can also be considered, and I think more animal studies are required to kind of understand the mechanistic pathways for both of these conditions. And these are my references, and I welcome any questions. Thank you. All right. Thank you. All right. Question. Well, yes. Thank you for a really impressive and beautiful study. Just a question about your exposure modeling. I saw that you modeled their exposure for the entire pregnancy. Did your exposure, was it based on extrapolation from monitoring sites in Southern California and then modeled for the entire pregnancy? Yes, it was. It was based on extrapolation from monitoring sites in Southern California and then modeled for the entire pregnancy. And I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. I'm not sure if that's true. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Great. Thank you. Thanks. All right. Winding on down here, we'll welcome our last speaker, who comes to us from the University of Cincinnati, and was it today, or is it tomorrow, that you're talking about Palestine? Is Palestine? Oh, we actually talked this morning. You talked this morning. Okay. So, I hope people made it to that. That was a quick add-on, as far as ACOM goes, and the residents had asked to try to move on for me here. But it was a commercial for something that's gone by already. So, welcome, Dr. Alexei, is it Krenyev? Krenyev. Okay. From the University of Cincinnati, who's going to talk on a different topic that probably Palestine got in the way of at some point. Hopefully, it'll poke us away, so to speak. So, it pulled you away from this there, but we'll hear about sharps and needle stick injuries in nurses and residents. Something near and dear to our hearts. Yes. All right. Good? Yep. Yep. I'm sorry. Thank you very much, Dr. Meyer. My name is Alexei Krenyev. I'm a second year OEM resident at the University of Cincinnati. And so, the topic of today's discussion is going to be an institutional study that we did on occupational sharps and needle stick injuries in the nursing and resident physician population and contributing factors and reporting and mitigation. To start out with, I'd like to acknowledge my funding sources. One is our University of Cincinnati NIOSH ERC, as well as the University of Cincinnati Department of Public Health Environmental Sciences Channing Meyer Fund. Likewise, personal disclosures, I have worked in a consultant role the last two years for marketing research. And our institutional review board approval was obtained last April through University of Cincinnati. So, just to give an overview, today we're going to give, we're going to go over the background and relevance of occupational sharps and needle stick injuries in the healthcare workforce. I'm going to give a brief project outline, discuss the retrospective and cross-sectional components, the conclusions, and where we hope to take our work in the future. Before I start, I would like to first acknowledge the invaluable help from my co-resident, Dr. Wally Changiri, and the support of Dr. Victoria Wilson, our program director. Likewise, I'd like to extend my thanks to Dr. Kermit Davis, our deputy ERC director, and then lastly, Dr. Rao, who helped with the statistical analysis. To give a background, a sharp or needle stick injury is really any type of skin-penetrating injury with a blood-contaminated sharp instrument or device. There are annually 6,000 to 800,000 such injuries in the U.S. healthcare workforce, according to the last CDC data. They exact a significant burden in terms of cost for the employer, one injury ranging cost from $71,000 to $3,000, with a total aggregate cost of close to half a billion dollars annually. Despite regulatory advances in terms of the OSHA blood-borne pathogen standard and the subsequent 2000 Needle Stick Safety and Prevention Act, there are, as I mentioned, close to half a million annually. The direct costs are significant and obvious. There are up to 20 known blood-borne pathogens with potential for percutaneous inoculation, and then also the potential for transmission of hepatitis B, C, and HIV, and also the indirect costs borne to the worker, including the anxiety, decreased productivity, and need for counseling. So just to give a quick background, there has been one of the questions we always kind of ask is, you know, exactly what's the scope of the problem? How many people have actually acquired HIV or hepatitis B or C from such an exposure? And the number is low, and thankfully, you know, two reasons. One is the introduction of antiviral drugs in the late 80s and 90s, and also universal hepatitis B vaccination. So between 1985 and 2013, from the data from the National Notifiable Disease Surveillance System, there have been 58 cases of HIV, 25 cases for hepatitis B, and 43 for hep C. And in terms of the broader scope, if we look at the data from the U.S. healthcare workforce, physicians and nurses really carry the greatest burden of injury. Forty-four percent of nurses and 20 percent of sharps and needle stick injuries recorded to the CDC through the National Surveillance System for healthcare workers were related to sharps and needle stick injuries. When we break this down in terms of what type of instrument was involved, leading to players are hollow-bore hypodermic needles and suture needles. With this in mind, we undertook a two-component study with a retrospective component and a cross-sectional anonymous survey to look at the prevalence, contributing factors, and under-reporting among the nursing and resident physician population at our medical center. We chose to focus on resident physicians as they occupy a unique role in their training, and also potential to impact with a future intervention to mitigate the effects. The retrospective study we evaluated recorded blood-borne pathogen exposures between November 2020 and October 2021. Of the 200 blood-borne pathogen exposures, 138 sharps and needle stick injuries were identified. Of those, 58 were among nurses and 64 were among residents. Currently, we initiated a cross-sectional survey by developing a 35-item anonymous survey using an expert panel of six members, one an industrial hygienist, two occupational nursing PhDs, and three occupational physicians. This was a front and back paper survey that was completed and placed in a white envelope and deposited in a metallic box. The components of the anonymous survey ranged from duration of work in the healthcare field to the type of instrument, to the resident specialty, and to the number of years of training in the healthcare field. In terms of the retrospective component, we used the injury reporting system, which the employees fill out for any work-related injury known as a ready set. We evaluated also the type of instrument involved and also the background regarding the worker. In terms of a cross-section of a retrospective study, as I mentioned, we looked at 138 sharps and needle stick injuries, and consistent with data from the CDC as well as the World Health Organization, at our institution, nurses and physicians carry the greatest burden. We next undertook to calculate rates of injury for nurses and resident physicians. Our denominator data was obtained through the Human Resources Department for the corresponding fiscal year, and so we calculated the number of sharps needle stick injuries per 100 residents. This was 11 per 100, and we did that similarly for nurses at 3.2 per 100. We next tested for statistical significance using a two-sample proportion test using R, and obtained a corresponding confidence interval, and residents had a rate threefold higher than nurses. We next proceeded to take the residents in a retrospective study and divide them into surgical and non-surgical categories. As we hypothesized that surgical specialties have a significantly higher rate than those that are in the non-surgical specialty. The non-surgical specialties range from internal medicine to dermatology, and the surgical specialties ranged from general surgery all the way to podiatry. We next, after dividing them into the two corresponding specialties, we one, did a two-by-two chi-square, where there was a significant difference in terms of the number of sharps and needle stick injuries recorded, and then we converted that into a rate, and also did a binomial test with a corresponding statistical significance. The surgical specialty residents had a rate almost tenfold higher. We then next broke down the rates per individual specialty and found that there was a significant difference among those in ENT, neurosurgery, OB-GYN, and general surgery, when the specialty-specific rate per 100 residents was compared to the overall aggregate rate at 11 sharps and needle stick per 100 residents. And also binomial exact test was used using R, and so we then proceeded to do a more descriptive study and to break down resident injury among instrument use. Among residents, suture needles followed by hypodermic needles accounted for most injuries. Among residents, likewise, there was a peak around 13 to 24 months, and also 49 months of injury. Resident physicians, likewise, had higher rates during a 12-hour shift, and the operating room, expectedly, was the site where most of resident injuries had occurred. We then proceeded to evaluate nurses, and with regard to nurses, the medical-surgical ward was the primary site. Hypodermic needles were the primary instrument, followed by suture needles. And there was a peak in nursing injury after 49 months of employment. In terms of shift block, nurses tended to have higher rates of injury in the morning and afternoon. Our cross-sectional study had, unfortunately, only 13 respondents from nurses, but we had a fair number of respondents from the residents. Among the 76 residents, and the survey was collected during resident service exams and also during monthly resident lunches, as many as 31 percent said that they never had a sharper needle stick injury, 26 percent really had no response, and 42 percent of the respondents said that they did have such injury. Among the 32 residents that had a positive history, there were a total of 74 recorded injuries during the course of their healthcare career. And if we plot that, we find that, really, most of that is in general surgery, followed by anesthesia. And then, when we do the same, when we look at the respondents that gave a positive history of injury, we find that most injuries occurred, like our retrospective study in the operating room. We then proceeded to evaluate the cumulative post-graduate year and the total number of sharps and needle stick injuries for that one specialty. When we removed an outlier from a resident that said they had a total of 10 in one year, we found that we had a weak negative relationship, so as cumulative post-graduate year improved, so as the residents got more experience, there was a corresponding fall. And in terms of shift, most occurred during the 12-hour shift. And then, we proceeded to look at non-reporting of sharps and needle stick injuries among resident physicians. As many as 41 percent had reported that they, during, at some point during their career, they had, that they failed to disclose a sharp or needle stick injury. Likewise, there was a significant number that said that they had recapped. And then, when we looked at recapping events over the last 12 months, there were several residents that recapped as many as 10 to 20 times a year, or one said that they had, they recapped up to 100 times. And then, the other thing we really asked was, were you fearful that you, that reporting the sharp or needle stick injury would reflect poorly on you? And as many as 16 percent of the respondents that had a previous sharp or needle stick injury said yes. So, the conclusions from our study is that, consistent with data from the CDC, as well as the World Health Organization, resident physicians really carry the greatest burden of sharps and needle stick injury in a healthcare setting. Those in a surgical subspecialty, in particular, carry the greatest burden, including ENT, OBGYN, neurosurgery, and general surgery. A high rate of non-reporting exists among the resident physician trainees, and their position as trainees influences their perception on how reporting influences their performance. Likewise, expectedly, as the cumulative training of the resident trainees increases, so is there a corresponding fall in the number of sharps and needle stick injuries. One of the conclusions is that this study really warrants, supports the need for further ergonomic study in the operating room environment to really delineate the specific drivers of injury, as the operating room environment where most of the injuries among residents occur is really a complex environment where there's an interplay of positioning and handoff of suture needles, and the drivers of injury are significantly different than on the floor or in the ED, where it's primarily hollow-bore needles. So one of the things we did is we constructed a Pareto chart using our retrospective data, and so most injuries among the resident physician population occurred during use, among surgical residents in the operating room, and due to suture needles. So one of the next steps that we hope to take is to develop a behavioral intervention using an appropriate health belief models to really address how surgical residents specifically perceive reporting of these injuries, and also, as I mentioned, an ergonomic analysis in the operating room to delineate specific drivers. Thank you very much. Thank you. Thanks so much. It looks like we have at least one question. Thanks very much. Impressive study. I was wondering, what are your thoughts about the differences between the specialties? I mean, it was huge differences among the different surgeons. Maybe with operating in the head region, what's the mechanism? Too little room, or? I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know. I don't know.

occupational medicine

biomonitoring

air pollution

hypertensive disorders of pregnancy

asbestos exposure

mesothelioma deaths

burn pit exposures

metal exposure

pulmonary function testing

cleaning and disinfecting

fibrosis perfusion

needlestick injuries

ergonomic factors