Symma Finn: Good afternoon, and welcome to the NIEHS Partnerships for Environmental Public Health Web Seminar, titled Mapping and Environmental Public Health, Visualizing Health Disparities. I’m Symma Finn, Program Administrator at NIEHS Division of Extramural Research and Training, and I’m charged with overseeing research in the social and behavioral sciences. I will be the Moderator for today’s session. We have four presenters for today – Dr. Alexander van Geen, Dr. Steven Chillrud, Ms. Meredith Golden, and Ms. Tricia Chai-Onn. The first presentation will be given by Dr. van Geen. Dr. van Geen’s research is focused on the geochemical cycling of trace elements in natural and mature environments, particularly [redocs] sensitive processes affecting metals and metalloids. As Associate Director of Columbia University’s Superfund Research Program since 2000 he has coordinated the efforts of an Interdisciplinary Team of earth, social, and public health scientists to reduce exposure of a sizable fraction of the rural population of Bangladesh to arsenic. This arsenic is contained in groundwater pumped from millions of private tube wells across the country. His presentation will focus on the importance of field kit testing wells for arsenic and mapping the results, of these results at the village level to identify the subset of aquifers that are low in arsenic. These aquifers are likely to remain the principal source of potable water within the affected areas for the foreseeable future. Dr. van Geen? Alexander van Geen: Thank you, very much, Symma, and organizers, of this webinar for this opportunity. I will try to make the point that mapping is, indeed, very important to identify an environmental health risk and in this particular case also to address it and mitigate a particular kind of exposure. The context that I will be covering, as you pointed out, is that of elevated arsenic concentrations in groundwater across south and southeast Asia. We have my colleagues at Lamont-Doherty and the Mailman School of Public Health and various institutions in Bangladesh have been working on this problem for the past 12 years, and I have selected some issues that are rather practical in nature to discuss today. The outline of the talk is as follows. I’ll show you what the distribution of the areas at risk is across the regions, say a few things about the regional patterns of arsenic contamination of groundwater within Bangladesh, and then switch mode to some extent and focus on what village level health workers can do with field kits and handheld GPS units. And I’ll show you some of these results, some of which were collected just over the past few weeks, and then point to some obstacles that are still present for reducing exposure further into the future. The map of – extending from Pakistan, all the way to China, that I’m showing in the third slide, is a map of population density. It is not an arsenic map, but the population density is shown for areas where for geological and climatological reasons one has reasons to be concerned about potentially elevated groundwater arsenic. And most of these regions have been tested, many of them have been found to be, indeed, to show, indeed, high arsenic levels in groundwater but not all. For instance, in Thailand there is an area that’s outlined, it turns out there’s no groundwater arsenic problem in Thailand. Integrating over the whole area, one of our colleagues, Peter Ravenscroft, has made a remarkable compilation and estimates that over 100 million people across south and southeast Asia and China are exposed to levels of arsenic that exceeds the current WHO guideline, which is 10 micrograms per liter. I will be referring, also, occasionally to the 50 microgram per liter arsenic, which is the drinking water standard in many of the affected countries and used to be until about a decade ago in the U.S., as well. Now why, first of all, do people rely so much on groundwater in this part of the world? Well, the image to the right is supposed to show you that sanitation is pretty simple, pretty basic in many of these densely populated rural areas. So the water in the pond, in the back of the picture, clearly that’s not the type of water that you’d want to drink. So that’s clearly one factor. Another one is that these [fluvial deltaic] areas that are outlined in the previous map, they are composed of unconsolidated sands and these sands are relatively easy to drill. The team to the left, the picture to the left, a group of three to four men, can install a tube completely by hand to 300 feet in depth within a single day. And the cost of a well like that will range between $50 to $300. PVC pipe is a key reason it’s relatively inexpensive to install these wells. For comparison, the per capita GNP of Bangladesh is on the order of $400. So it is clearly a significant investment. The fact that millions of people across the country and across the regions has done so, shows to what extent safe drinking water is valued. The picture to the right also shows what these pumps look like. It’s a very basic cast iron hand pump placed on top of a string of PVC pipe, with some metal pipe towards the end, and also a concrete platform built around it. Switching to the next slide is a map of Bangladesh, summarizing the most extensive survey that has been conducted in the country to date. The colored areas are based on field kit tests for arsenic done for – carried out for five million wells, an enormous number. And the areas indicate the severity of the problem, so the light blue areas, for instance, less than 20% of the wells were found to exceed the 50 microgram per liter standard in Bangladesh, and the red ones more than 80% of the wells were exceeding that. So a first mapping problem, if you will. If you want to address that there’s clearly more urgency in the red areas than there is in the untested areas, the black areas, for instance because those had been tested although not at the same scale. So a first point about mapping environmental health risk. The yellow box that I’m showing you is where my colleagues at Columbia University and in Bangladesh have been conducting a cohort study, started this covert study in 2000 and it is still ongoing, with supports primarily from the Superfund Research Program of NIEHS. Zooming into a considerable portion of this area is the next map, where every individual dot is – corresponds to the location of a tube well, like the one I showed you in the first picture, and each of these wells a sample was collected and was tested for arsenic in the lab, initially, by [graphite furnaces], a, atomic absorption, and eventually by [ICPMS], as well. What you can see is the clusters of wells, which correspond to the location of the villages, following, if you will, the contours of where a major river runs today or following contours where rivers used to be, so the villages are typically located on slightly elevated areas, riverbanks, in particular. So what you can see is tremendous variability of arsenic levels. Some villages entirely low in arsenic, the light blue dots, of course, indicate wells that meet the individual guideline, the green ones the Bangladesh standard, and the ones that are red all exceed the 50 microgram per liter arsenic. I’m a geoscientist, I’m not a specialist in public health, although hopefully all of us involved with Mailman have learned a few things. I would just summarize a few key features of some of our main findings so far, but there are many others. If you look at the cohort of about 11,000 people who were recruited back in 2000, 2001, there have been on the order of 400 to 500 deaths. And if you compare the number of deaths for the high exposure group compared to the lower exposure group, in this case we’re using 10 and 150 microgram per liter as some defined definitions of these groups, the people in the high exposure group over a [nine] year period were twice as likely to die. So it’s an enormous health impact. What came, I believe, as somewhat of a surprise is that most of these deaths actually came from the consequences of cardiovascular disease. Arsenic exposure has other consequences, like cancers, and our team has also found it had impact on the mental and motor function of children drinking the high arsenic water. But the most dominant effect, the most striking effect clearly is the doubling of death rates from all causes, clearly a significant health problem. Now I’m going to zoom in to make several more points about mapping environmental health risk. I’m going to zoom in on one particular village, called Edbardi Village, which is located in the southern portion of the area. I’ve outlined it in the yellow box, again. And if you go to the next slide you have the data collected. In this case I’m going to show you actually data that was collected just a month ago by a single woman from a village, from that particular village, using field kits and handheld GPS receivers. Now you will see, just to point out a few features of the village, you have the stream, you have a road, and then the village containing on the order of 350 wells, typically 10 to 15 people drinking from each well, to families, so we have a village of about 3,000 inhabitants here. You see, again, that even when you zoom in at the village level the highly variable distribution of some low arsenic wells, some high arsenic wells, and some intermediate levels, as well. Overall in this village half the wells are – don’t meet the Bangladesh standards. A few more words about how the tests were conducted. Here is the particular lady who conducted this test. She is holding a handheld GPS in her hands, and that is turning out to be very useful because it’s not a particularly sophisticated unit, but it’s a unit that allows the tester to install, to enter enough characters into the GPS and tie to the GPS coordinates so that the information can be downloaded. And there’s also a backup, a paper backup, and here’s the little kit that is being used to do the testing. The reaction time of this particular kit is on the order of 10 minutes. We are running this project with 10 village health workers right now, with two supervisors, and including the cost of all the salaries, all the materials, I can tell you that it’s less than $2 per test, and we’re keeping track of that because we think this program should be expanded throughout the region. So with the results of the test the wells are identified, they’re unsafe, and the safe ones identified in the three categories I mentioned earlier based on the test results, it’s a scale in milligrams per liter below. And so either light blue or green or a red placard is placed on the well. It turns out that about a third of the cost per test is the making of these stainless steel placards. Now let me point out a first form of mitigation that actually we all underestimated initially. It turns out that in our study area, and I’m zooming here in on the portion of the village, and you can see there’s a length scale here of 200 feet, in the lower left, on the Google Earth image, 60 meters. And essentially what it shows you is that most households live within walking distance of a safe well, even if their own well is unsafe, and has one of these red inverted drops. When we carried out this calculation of our whole study area, and it turns out that even though half the wells in our study area are – don’t meet the Bangladesh standard, 90% of them actually live within 100 meters of a safe well and could potentially switch to a safe well. It turns out although some of our colleagues in Bangladesh thought that not much of this would happen, it turns out that over the years half, more than half the people who should do something about it have actually switched to a neighboring well. So an encouraging result, although not the 100%, of course, that we are seeking. The next slide shows you the depth distribution of the wells, as well as when they were installed, because that’s one of the questions that our village health workers ask the households. And you can see a couple of things, mapping, if you will, in a vertical dimension, as well. First, you can see that there is a mix of high and low arsenic wells in the upper 100 feet, and so clearly if you want to install a new well you don’t want to be installing within that integral. But if you were to install beyond, say, 120 feet you can see that almost all the wells and there could be some data errors, maybe some drills we didn’t quite install as deeply as they should. But you can see that by and large all these wells are low, very low in arsenic, and that reflects the geological history of the deeper deposits. And in many villages these kinds of traditions are observed, although the depths of these transitions will vary depending on the local geology from one village to the other. So installing deeper wells is key – Symma Finn: Dr. van Geen? Alexander van Geen: Yes? Symma Finn: Two minutes, Dr. van Geen. Alexander van Geen: Yes, I’m going to wrap up within that time. So this is mitigation option number two. Now I don’t want to leave you with the impression that everything has been solved. I’ve shown you earlier a map showing the data that was obtained in 2012. Before this was done, it turns out that as these yellow drops indicate, a lot of wells, essentially half the wells in the village had not been tested, and that’s because they were installed after our testing campaign, as well as a government campaign, testing campaign that was carried out in 2003. So that’s one thing, and then the little Y’s in this Google Earth image indicate households that know that their well is unsafe and yet are not doing something about it, and it’s about 50% of these people. So clearly there’s also something related to risk awareness that needs to be emphasized. So to wrap up using the same outline I had as my second slide, I’ve tried to show you that the problem extends from Pakistan, all the way to Vietnam and to China, over 100 million people affected. There are regional patterns that help – that needs to be – that are known and are useful for targeting mitigation. I’ve mentioned our main sort of health result from the cohort study. I hope I’ve convinced you that you don’t need a laboratory, you don’t need a very expensive intervention to test these wells, and testing these wells is really the main – the most urgent thing that needs to be done at this point because wells continue to be installed. And for that reason we are now exploring the possibility of instituting a commercial testing service at the village level. Most of the wells are privately commercially installed. It seems that testing, which is supposed to be done by the government isn’t taking place. I think this could be addressed with a commercial service, and maybe us in academia, our task is to come up with a certification program that can be – that is credible and a training program to favor such commercial testing service. And I’ll leave it at that. Symma Finn: Thank you very much, Dr. van Geen. A most excellent presentation. We had a question here regarding the field kit testing by village health workers. You showed a picture of a local person. How many have been trained, and who does that training? Alexander van Geen: So in the country scale testing campaign that was done back in 2003, hundreds of NGO workers had been hired, and I think they had a higher education level than this particular lady who has completed primary education only. We have tested 10 of them. Our Public Health Team has many more, village health workers, you know, carrying out sampling, urine sample collection, than with doctors, they collect blood samples. These are the people that we – that live in the villages, they know they have untested wells – we think that these are the people that should be trained rather than outsiders. Symma Finn: Thank you. We do have – we have received a number of questions. We’ll try to take as many as we can in the remaining few minutes. The first question was are the well tests reproducible over time? Can a low arsenic well become a high arsenic well over time? Alexander van Geen: That’s a very good question. And within our field there was a lot of concern about that. The monitoring that has been carried out over now more than a decade shows that, by and large, arsenic concentrations don’t change over time, so there’s considerable spatial variability but there isn’t the same level of temporal variability. That said, there are some wells that are more vulnerable than others, and when the arsenic concentrations do change it typically happens gradually because of changes in groundwater flow patterns, irrigation pumping is carried out on a large scale, but it is very gradual. If it happens suddenly then it’s more because of the mechanical failure, in fact, out of the hundreds community deeper, community wells that we’ve installed, a handful, half filled, and in most cases it’s actually because the PVC pipe became disconnected at the shallow interval and so some groundwater could be coming in for that reason. Symma Finn: Thank you. Another question was have your findings on the depth of the wells and the arsenic concentration informed current well drilling or is the cost difference a problem? Alexander van Geen: The village I showed you, we know that the driller we work with for our research has been busy there, and I think that’s why it’s a pretty good illustration of how local drills can use this information. We’ve actually made a calculation of this type for the whole country for over 40,000 villages with where that transition is and how robust that estimate is based on the distribution. Unfortunately, this has been one of our main disappointments in the project for reasons we still don’t quite understand, and even though we had set up an agreement with a main mobile phone provider, Grameenphone in Bangladesh, we have not been able to distribute this information throughout the country. Symma Finn: We have time for just one more question, and thanks to all who are submitting them. The question is do the field workers enter data into Google Earth or is this done by the University? Alexander van Geen: So, good question. So they enter the data in a sort of format, fixed format in the GPS units. The supervision level, I have two supervisors in the field, and he can handle the 10 village health workers. He has a local college degree. He knows how to connect the GPS through his laptop and can upload the data into first Excel and then Google Earth Pro. So that is the level of skill that is needed. The village health worker doesn’t need to do that herself. I am encouraging our supervisors to produce the maps of the type that I’ve shown you. Symma Finn: Thank you, again, Dr. van Geen. We have received a number of other questions, which we will forward to the presenters after this webinar for them to respond directly. At this time, I would like to introduce our second presenter, Dr. Steven Chillrud. Dr. Chillrud is a Research Professor at Lamont-Doherty Earth Observatory and Director of the Exposure Assessment Facility Corps of Columbia University’s Center for Environmental Health in Northern Manhattan. Much of his research is focused on the role of particles in the transport, behavior, and fate of chemical contaminants. His air pollution work includes understanding the sources, behavior, and exposure pathways of airborne contaminants, as well as designing and testing new air monitoring devices, either to be used at fixed indoor and outdoor locations or to be worn by people to understand their personal exposure. He works in collaboration with public health investigators on a range of local and international health issues. Dr. Chillrud? Steven Chillrud: Thank you, Symma. It’s a real pleasure to present today, and I had originally presented or prepared a much longer talk and, but so today we’re just going to be giving highlights of that, examining how geospatial analysis can help lead to a better understanding of sources and exposure pathways. And this has been done with the help of and in collaboration with many others, which we’ll see at the end. So my outline is – I’m going to start off giving an introduction of the spatial and temporal variability in air pollution in New York City, focusing on ambient metals with an overview. This came from our study from the late 1990s where Pat Kennedy was the lead PI. Then I’m going to go on and the first time that I used GIS mapping and what it did was it gave us a hint to understand what the exposure pathway was for these New York City adolescents for three different metals that we measured, a very simple application. And then I’m going to go to the Neighborhood Asthma & Allergy Study, we have a very recent work, where we use the ArcGIS functions, and again in a very simple way to help identify residential sources of soot or black carbon getting into the homes and in relationship to asthma, early markers of asthma there. And then if I have time I’ll look at some personal monitoring results in GPS with the real-time monitors, so someone will probably cut me off because I probably don’t have time. So the TEACH Study was funded in the late 1990s, directly out of the Clean Air Act Amendments of 1990, which basically said that in occupational settings where levels of air pollutants could be extremely high relative to ambient levels there were known health effects, but very little was known in the general population which might have susceptible subpopulations of what the exposure pathways or the levels were of a wide range of hazardous air pollutants. And so we got funded to do that. And our study design looked at the two different cities and two different seasons, we’re looking at different types of sources. But the main study design was really focused to try to understand the spatial and temporal variability so that we could understand local and regional sources and how that affected the personal exposures. And it was based on having an urban central site that was measured three times a week for 48 hours, an upwind site that was representative of the upwind air masses, and then this subject based sampling where we did five subjects a week for the same 48-hour period, where we measured home indoors, home outdoors, and a personal sample in a backpack that the students carried around. And we were measuring a wide array of VOCs, and aldehydes, but also collecting fine particulate matter and measuring the black carbon and the elements on those filters. And so here’s an example from the summer for arsenic, air concentration on the Y axis time, on the X axis, and there’s actually five homes in pink shown here, that home outdoor sample in this case, and in addition there’s the three samples per week for the fixed sites with the yellow being the upwind fixed site and the blue squares being the urban fixed site. And so what you see in this plot is that there’s very little spatial variability shown for the samples that are collected at the same times, and there’s not much difference between the upwind fixed site and the urban fixed site. And so the simple interpretation of this is that arsenic comes from the upwind regional sources, and it’s been known for some time that the peak demand in electricity production in the Ohio Valley puts out arsenic and that’s where the vast majority of the arsenic in New York City comes from in the summer. We can compare that to cobalt, this time in the winter, where there’s a large gradient going from the upwind site to the urban site, and there’s a lot of spatial variability for the samples that are collected for the same 48-hour period. And so here the clear interpretation is that cobalt has many local urban sources and so you can do that for all the elements, and we can actually kind of summarize them looking at this plot here, where we’re just taking the median of the daily ratios of the urban fixed site, the upwind fixed site, and then ranking them in order from a low ratio to the high ratio. And so that way elements that plot on the left-hand side of this with a ratio near one are the ones that are coming from the upwind air mass. You can see selenium and arsenic there, and a bunch of other elements, where elements that fall on the right-hand side that have the elevated levels and much more elevated in the winter, like cobalt, lanthanum, and nickel, are coming from local sources and here largely from the use of fuel oil to heat the homes, and as well as one of the big fuel elements, vanadium falls in the middle, so there’s probably multiple sources of that both coming from upwind and local sources. And so that gives kind of a nice overview of the spatial and temporal – oh, let me just go back one plot – the other thing we see here is in the temporal is that New York City, like much of the northeast, has weekly weather patterns. And so everything goes up and down on this kind of six to seven-day cycle, and that drives a lot of the variability and time. So the next thing, so I’m going to talk a little bit more about the TEACH Study because one of the big findings we saw was that certain metals, the personal sample from the backpack that the students wore for 48 hours, had a distribution that looked very different from all the other types of samples we took, that didn’t look anything like what we sampled indoors, outdoors, the urban fixed site or the upwind fixed site. And we spent a lot of time trying to figure out where this was coming from. Early on we were very suspicious of metal that was accreting from our personal backpack or from the metal [cyclone] that was – but we ruled that out, and the correlations between other metals – this is true, this basic picture was true for iron, magnesium, chromium, and those three metals were highly correlated in all the samples but you had a very different ratio in the personal samples versus all the others, all the other locations’ samples plotted on our ratio about half of that for the personal samples, indicating a single source. And so the next thing we did was we actually plotted for the home locations, for the personal samples, who had elevated concentrations and elevated ratios versus who had moderate or low concentrations. And in the winter everyone who lived far from the school that they had to commute to had elevated concentrations and elevated ratios, or in the summer when they weren’t going to school you had a more mixed bag of what was going on. And so, to us, that was a clear hint that the people who lived far away were getting this on their commute to the school, and the way you commute in New York City if you have to get anywhere in a reasonable amount of time is by subway. So we sent a summer or undergraduate thesis student down into the subway to collect an eight-hour sample while they were just on the subway platforms or riding the subway cars, and those samples were totally consistent with these elevated and distinct ratios that we saw in the personal samples. And so that’s just a nice snapshot of really how we got to that conclusion. So our next highlight comes from the New York City Neighborhood Asthma and Allergy Study. This is a study that [Matt Prosonowski] leads. It’s a cross-sectional case control study of asthma among seven-year-old children living throughout New York City. The recruitment is all done based on a middle income health insurance plan to try to make the different groups more similar in SES. And the neighborhoods were selected based on the New York City Department of Health asthma prevalence among five-year-olds, with half of the cohort coming from lower asthma prevalence neighborhoods where the asthma prevalence was between 3% and 9% and half came from high asthma prevalence neighborhoods of 11% to 19%. And during the home visits allergens were measured in the house dust, serum samples were collected to measure IGE antibodies, detailed symptom history was collected, and a seven-day residential air monitor was set up and collected fine particulate matter, and then black carbon was measured on that with an optical method, as well as some biomarkers were collected from exhaled breath. And so here is where all these different kids lived, where the 347 families lived, and some of them are in the higher asthma prevalence neighborhoods, some of them are in lower asthma prevalence neighborhoods, so black carbon, which is a marker of incomplete combustion byproducts and has often been used as a tracer of diesel emissions, was on average lower, in both low asthma prevalence neighborhoods, and higher in the high asthma prevalence neighborhoods. And to look, to think of that, the typical thing that many groups would do would be just to look at traffic density, but we also because of recent release of data from the Department of Health, the Environmental Defense Fund did a very nice job releasing to the public and actually having a nice mapping program of its own, where you can see exactly where this residual fuel oil, number four and number six, is burned in about 10,000 apartment buildings in New York City. And it’s – this residual fuel oil is much dirtier and it leads to a lot of incomplete combustion. We used that data to basically do a very simple analysis where we’re using ArcGIS function, where you put a buffer around the home location and you just count up the number of homes that were burning the residual fuel oil. Steven Chillrud: So when you do that and you – and when you plot the buildings burning within that 500 buffer number, ranging from zero up to more than 120, versus black carbon, you get a positive association just like you do if you look at the truck route density. And so this is just a nice example of a simple thing that you can do in ArcGIS, and you can also do it in Google Map, it’s just you have to count it by hand. Steven Chillrud: So the other thing we did there is we looked at a marker of fractional exhaled nitric oxide as a marker of inflammation, as an early indicator of what might lead to an asthma exacerbation, and again, and then what we saw was depending on the [serioatypi], whether someone was – had IGE antibodies in their blood, showing that they were allergic to allergens, those people who were did not show a relationship between airborne black carbon and this marker of inflammation or those who were not did. And so that was a nice outcome from this. And basically it’s another piece of information that is defending Mayor Bloomberg’s recent legislation, the Department of Health in New York City [VDP] where they’re trying to phase out this residual oil that is so dirty. And they’ve gone through in a number of statements or a number of local and state laws to phase out this use of this oil by 2013 in the end. And it’s mainly by trying to replace boilers, they’re trying to switch them to other heating fuel that can still burn, the viscosity of the oil is an issue so you can’t just jump from the number six all the way to number two. You actually have to replace the boilers, and because of that expense there’s beginning to be quite a bit of pushback, and so we would argue that they should not let it happen. But, of course, once you get to a boiler that can burn number two fuel oil you can actually do much better than number two. So here’s the sulfur content of fuel oils. The number six has 30,000 [PPM] sulfur and it’s the sulfur content that helps make it burn so dirty. Number four has 3,000 or more PPM sulfur. The traditional fuel, number two, has 2,000. The lower sulfur diesel has 500, but the ultra-low sulfur diesel that’s being burned has only 15 PPM. So once you can burn number two, it’s a clear pathway to actually doing much better than the current legislation. Steven Chillrud: And that’s where I’ll end. Symma Finn: Thank you so very much for a most interesting presentation. I’m going to wait a moment and see if we receive any questions, and in the meantime ask whether you considered socioeconomics as an additional variable when you were looking at – because I notice you use the same SES, but to track health disparities whether you might introduce another variable? Steven Chillrud: Well, now, they did correct for – I forget which variable they used for SES, but they did correct for a large number of variables, including SES, and this is a paper that actually just came out and people can look and find it, it’s the [Cornell, et al] paper that was just released, it’s out in the journal form now. Symma Finn: We do have a question here, how do you separate out the role of allergens and allergen sensitization to your analysis? Steven Chillrud: Well, the way that Dr. Prozonowski did it was he measured IGE in the serum, and so if IGE was above a certain threshold level they were considered to be sensitized to the allergens and were divided into one sub group, and if they – if their IGE, and this was to cockroach allergen, mouse allergen, and dust mite allergen, I believe, was the three that they looked at – and if they were below a threshold then they were considered to be non-atopic and there was a clear – and by doing this they saw this clear different response to this marker of inflammation, this fractional exhaled. Symma Finn: We did – we have another question for you – were the results of the personal versus external measurements of contaminant similar in terms to the type of contaminant? Or were certain metals higher than others? They noticed that you showed results for iron. Steven Chillrud: Right, so for – there were three metals that were related to this subway pathway of exposure from routing the subways during their short commutes to school, and those were iron, manganese, and chromium, and those clearly indicated a source of steel dust. And so when the subways were being or growing the wheels get upgraded and that steel dust gets into the air, and some of these kids only commute for 15 minutes to get to school but they still show these much higher levels. There were a number of other metals that are also elevated in the subway related to some of the electrical contacts and brake linings or rat poison, but not nearly as much as the main steel dust elements that are of the iron, manganese and chromium. Symma Finn: And one final question, did the students map their movements? Steven Chillrud: So, not back then in the late ’90s, but the one that I didn’t get to was the more recent work where we’re using real-time data. And so they’re wearing a GPS data logger, you know, a real logger that costs about $100 and it’s just a couple ounces. And so here’s three tracks from three different subjects that wore that. The one in yellow is actually a child that gets driven to school, lives up here in the Bronx but gets driven to school by his parents down to a school that’s down here. And the clear thing we see is during that early morning commute is incredibly high levels of PM 2.5 in the real-time marker because the parent actually smokes in the car and because of the small space and the limited air exchange, you get very high levels of the cigarette smoke exposure in that setting, and so we see that the speed of the GPS and the timing of the GPS lining up with this real-time marker that we were able to identify the in-car smoking behavior as the cause of that high exposure. Symma Finn: Thank you so much. We’re going to go on to our third presenters. At this time I would like to introduce Meredith Golden and her colleague, Tricia Chai-Onn. Meredith Golden is a Senior Research Associate with the Center for International Earth Science Information Network at the Columbia University Earth Institute. Her academic background is in medical geography, epidemiology, and urban and regional planning. Meredith is the co-PI of the Research Translation Core for Columbia’s NIEHS Superfund Research Program on the Health Effects and Geochemistry of Arsenic and Manganese, for which she coordinates outreach, translation, and geospatial approaches for Superfund Research and activities. Over the past 30 years she has worked on a variety of issues related to the development and implementation of Superfund policies and regulations. Tricia Chai-Onn is a Geographic Information Specialist at the Earth Institute Center for International Earth Science Information Network. She received her Masters in Urban Spatial Analytics from the University of Pennsylvania and her Bachelors in Psychology with a Minor in Environmental Science from Barnard College. Her key interests are in cartography and public health. As part of the Columbia University Superfund Research Program, Tricia has worked on the assessment of populations in proximity to Superfund National Priorities List Sites Report and the creation of the NPL Superfund Footprint Mapper. Both projects have required her GIS skills and her expertise in identifying, acquiring, and managing [demographic?], environmental and health data. Ms. Golden and Ms. Chai-Onn? Meredith Golden: Thank you very much, Symma. Tricia and I welcome the opportunity to showcase the recently released NPL Superfund Footprint Mapper. Unlike the other talks that you have heard today this one is not about research per se, but about a tool that we hope will guide and translate research through visualization. The Columbia Superfund Research Program Research Translation Core received a supplement through its P42 Grant to develop an interactive online mapping tool to help researchers, practitioners, and regulators and the general public better understand the characteristics of vulnerable populations, build the natural features and sources and houses exposures near Superfund sites. Its development was led by our Interdisciplinary Team, the Research Translation Core, scientists at CIESIN. CIESIN is part of the Columbia Earth Institute’s Center for International Earth Science Information Network. Here you see the image of the Superfund Footprint Mapper. It can be accessed via the URL given on the top or directly from Columbia Superfund Research Program website. The Mapper opens to show this image with the NPL sites giving a one-mile buffer. Across the U.S. there are a total of 1,716 NPL Superfund sites, over 1,500 are shown as polygons, and the remainder as points, about 145 don’t have a footprint yet. Maps are useful media for visualizing and thus translating information to a broad audience regarding the geography, population, physical and built environments over space. Here’s a quote by [Kim] Elmore, who might be in our audience today, and her colleagues, on the application of geospatial approaches to public health in terms of priorities and strategies from an expert panel convened by the CDC in 2008. It reads, with the democratization of data tools and techniques for basic geospatial analysis and cartography, along with gains in conceptualizing the role of place in public health have provided an empirical evidence of geographic difference in a variety of health related outcomes. When I was first studying epidemiology in the early 1980s at Chapel Hill very few researchers considered any spatial risk factors. Now, however, it’s widely accepted that place impacts health and health status varies over space. Using maps is one way to explore the geospatial dynamics of public health issues. Here are a few good reasons to use GIS as part of environmental public health research and outreach. Its ability to sort and integrate data from multiple sources, to utilize new types of data from GPS and satellites, its usefulness as an epitool for exploratory analysis. It’s also an extremely effective tool for communicating a link between environmental exposures and public health concerns. These pictures of Barber Orchard Superfund Site in Waynesville, North Carolina depict how over time different land uses can bring toxic exposures close to home. Maps can show the layers of these exposures and the proximity of populations. Here’s some of the reasons that we decided to develop an online interactive mapping service as part of our Research Translation Core. Maps by themselves are good, but a mapping service offers so much more. You can have quick visualization, easy access to data on NPL sites, multiple hazardous exposures and vulnerable populations. You can choose which scale to look at, whether at the national scale or regional scale or zooming into a local scale. Also, it can help translate and communicate information for intervention, cleanups, community engagement and many other purposes. Now the mechanics of this, the Superfund Mapper is built on the Esri web application development framework or WebADF. It integrates data layers and arc map and the map document is published as a service. The ArcGIS online topographic base map is also pulled into the application as a service. And although we didn’t use it, it would be possible to use Google Earth as the base map for this. What we did was we created one and four-mile buffers around the NPL points and the ATSDR polygons. As you can see here, on the left is the point data that EPA provides, the centroid of the site, and on the right is an example of the polygon that ATSDR has gone to each of about 1,500 sites and actually plotted. And then you can basically draw these buffers on the map. And the buffer is measured from, as I said, from a centroid from the left and from the edge or the border of the polygon on the right. Having sites represented as polygons obviously give a more realistic picture of what’s in proximity of the NPL site. This is probably a more dramatic example would be, which I’m not showing here, is of the Hudson River, which is a Superfund site that is over 200 miles long. If you went from the centroid of that you would not capture the population that is within a mile of the border of the entire length of that 200 miles. So we felt in creating this mapping service one of the big services that we were doing was making accessible the polygon information. To determine the population numbers within the buffers around the NPL sites we implemented the majority rule, which reduces overestimation. If 50% or more of the gridded cells fall within the buffer, the entire value is assigned to that cell. This is done for each of the selected variables, demographic variables, and included in the calculations. Here are just some of the layers that we’ve included in the Superfund Footprint Mapper. Polygons, as I just mentioned, with the actual Superfund site footprint, 30 socioeconomic variables for populations within the one and four-mile buffers, information on public and private schools and colleges, their location, their names, and the number of students, brownfields, TRI data, high-priority chemicals, landslide probabilities, fault lines, population densities, infant mortality rates by race, education, social vulnerability index, watersheds, wetlands, and the footprints of American Indian and other indigenous people, they’re basically the areas where they live. So why do we think that the Superfund Footprint Mapper is useful? Well, we worked very hard to include information that would be valuable to a variety of users and functions that would be easy to use for everyone. So here if you click on quick tour, which is in the menu above the map, you can go directly to a series of documents. We’re just showing one here, which is the quick tour, itself, but we have included a series of documents describing how to use all the components of the Mapper plus useful citation information. We also have embedded the legends in the map contents. So, again, you can see to the right of the map is about contents menu. You click on the plus side to the left of each feature and the legend for that feature appears. For example, you see the green triangle for TRI locations, the aqua borders for EPA regions, and a list of values corresponding to the U.S. population density estimates in terms of persons per square kilometer. We’ve also provided metadata for each of the data layers. So, again, if you clicked on the metadata guide, you will come up with a document that has each of the data layer names, and then it has the source of the data, the year that the data relates to, the description of the data, the geographical unit, such as area or point, whether the data on the map is scale dependent. For example, in some cases we don’t include wetlands at the national scale. You need to basically zoom in before you’ll see it and so, therefore, it’s muted in the map contents on the left. And we also provide links to the actual data or to the data sources and/or the data sources. The Mapper has multiple tools for zooming. In addition to the slider, that you can see on the left of the map, you can also use your mouse and do a scroll bar to zoom in or out, and then you can also click on zoom to EPA button at the top of the menu. A box appears on the map, as you can see on this slide, with all the EPA regions to choose from. For EPA Regions 2, 9, and 10, they all have two options – you can – this is so that you can see more easily Puerto Rico, Hawaii, and Alaska in their respective regions of 2, 9, and 10. In this case we have zoomed into EPA Region 4. You can notice that there’s an aqua border that delineates the states within the region and the number four is in the center of the region. We’ve included, also, an identifier, an identify pointer, so users can click anywhere on the map and access data for that location. This is a little complicated slide, but it shows how both the magnifier and the identifier pointer in the menu above the map have been zoomed first to zoom into the area near the Ringwood, New Jersey Superfund site, which is a magenta figure on the map, and the actual footprint of that site, along with the one-mile buffer, which is the light – is also a magenta border. The light yellow outline, that I hope you can see, is – shows the land that the [Ramapo Indian Tribe] occupies. And, as you can see the Superfund site happens to be almost entirely within their land area. The identifier box that is shown, I had used the I and clicked on it, and it gives you a lot of features for the point where I clicked. And, as you can read, there are wetlands, it has NPL site buffer at one mile, which later you’ll see gives you a lot of demographic data. The NPL site will link you to all the CERCLIS data. And then the Ramapo area, if you had clicked on that, you would basically get the demographics. I realize we need to step this up just a little bit. Symma Finn: Yes, two minutes, Meredith. Meredith Golden: Actually, you can save your results. Here we’ve identified a school within one mile of the site. It shows that there are 305 students there. This information, along with the results from the other features, can be saved. And you look on the results section on the upper left from the map, and you can open any of these tabs, actually, to see the data within them. I’m not going to show that right now. But what you can do, too, is you can print the map and the data results. This is what it looks like when you go up to the far left corner and you click on the print, you get the box that opens, you can label your results, you can decide what size map, the quality, whether you only want to print the results or the results plus the data. In this case you can also save, it is a PDF, in all cases you can save it as a PDF. This is the first page that shows the map that you had just created with the legend for that map, and here is an example of the results for that. I circled women of age 18 to 44, that’s childbearing age, because to me that number looked low. I made our people check and it was not an error, so that immediately triggers in my mind that this is something I want to look into and see whether those women have migrated out in that age group or if there’s a health concern. So this is a type of a way that information that you get can trigger some future research. Finally, it’s possible to add new layers. In this case I have included a photo from a very interesting site, called Superfund 365, that has pictures from many of the sites. You can add in, you could basically link to this within the Mapper, we haven’t done that yet, but you can add photos, videos, stories from the localities, other data layers from high resolution data that maybe some of the other researchers or government officials have for the data sites. So moving forward under our renewed SRP Grant we are planning to update the population data and work with ATSDR to create polygons for the remaining NPL sites and we’ll have point data, with more funding or through collaborations we’d also like to add more data, especially health data, expand our zoom capabilities to zoom into states, counties, or specific sites. We really would like to add query tools and some basic spatial analysis techniques to this. But most of all we’d like to hear from you what you think would be useful for what you would be doing regarding the Superfund sites and let us know how we can collaborate together in the future. I want to give special thanks, first to Tricia Chai-Onn, who has assisted in this presentation and led the Geospatial Team. Acknowledgements are, again, that the project was funded by the Superfund Research Program as a supplemental grant. Our Team included myself and Tricia, Kytt McManus, Greg Yetman, James Carcone, Annie Gerard, and I’d like to give special thanks for the support and data. When I came up with this idea I wasn’t sure just how we were going to implement it, but we were very lucky in that Andy Dent at ATSDR GRASP gave us the polygon data and that we could use, and Randy Hippen, with EPA, provided us with the data for the points. Bill Suk, Beth Anderson, Danielle Carlin, and Mary Gant all were supportive in helping us make this a reality. And, again, also Larry Reed, Rebecca Wilson, and Maureen Avarkian helped us along the way. And thanks, too, to the numerous beta testers, but at the same time I feel we, even though it’s in production we do want feedback from all of you. I hope you’ll have a chance to go use it. Here is the URL again, and there is a button for your feedback, more than welcome. So thank you very much, and if you have any questions just let me know. Symma Finn: Thank you so much. A most interesting presentation. We have already received our first question, which is what socioeconomic data sources were used for the population characteristics? Meredith Golden: Okay, for the demographic population we basically here at CIESIN have taken the U.S. Census data and gridded the population. This is – was done under our NASA grant for our Socioeconomic Data and Applications Center. You can find the specifics of all of that in the metadata Mapper, but some of the population data that we did, for example, on high school students came from the DHHS Community Health Status Indicators Project, and the infant mortality rates were also part of a CIESIN project that we had – our demographers had put that data together. So it’s somewhat of a variety of sources, but most of the demographic data was from the U.S. Census grids that CIESIN has taken the U.S. Census data and then gridded it. Symma Finn: Thank you. Since health data is mostly available at the zip code and census track data how will you address adding these layers to your data without introducing uncertainties in trying to combine all of these data available on different geographic resolutions? Meredith Golden: Well, that – it is a real test to try to make sure that the data is in a format that you can integrate it. I’m sorry, Tricia, did you want to say something about this maybe – no? So that’s a kind of complicated question, which I am not sure immediately of the answer to, but what I would like to do is whoever submitted that we could sit and talk with our demographers and figure out what would be the best way to integrate that. So, sorry, that was kind of a non-answer from me. Symma Finn: We have two related questions. They are who do you see as the primary user of this tool or is it primarily a tool for you? And the other question was is the grid’s population data available to the public? Meredith Golden: All of the data used is available to the public, especially CIESIN is as a Center generates and integrates a lot of data and our bottom line is that it has to be freely available and accessible to the public, so, yes, all the data is available to the public. Some – we have modified some of the data in order to integrate it, and by that I just mean perhaps it was not – I’m thinking of some of the CERCLIS data we’ve gone through and matched according to the EPA IDs, et cetera, and we have those data sets, the integrated data sets also available. So in some cases we have the primary data and in other cases we also have the – basically the data that we have cleaned up and managed for this Mapper also available. I’m sorry, I didn’t quite understand the first question, though? Symma Finn: The first question is who do you see as the primary user of this tool or is it primarily a tool for you? Meredith Golden: Yes, I see. Well, we’re hoping that our scientists will use it, but actually I’m hoping that to start with that it would be useful to our EPA partners. They have some other Mappers but the other Mappers don’t focus, per se, on Superfund and they don’t have the footprint, they don’t have the polygons, which I think is a very important aspect of it. So I’m hoping that it will be useful to people at EPA that work with the communities around the Superfund site, either as their community involvement teams or as their remediation teams, and might also be useful – and I’m sure I’ll get feedback I hope from ATSDR to see if this helps with their assessment of Superfund sites. And, with that in mind, is that for the Superfund Research Program there are I believe about 18 universities that are part of the program, receiving grants, and I’m hoping both that they can use this for their own research at Superfund sites and also that they will then give us some data layers from what they’re collecting that we can integrate. Symma Finn: Thank you. Are there any other questions for either Dr. Chillrud or Meredith and Tricia? Unfortunately, Dr. van Geen had to leave for another meeting and is not here to answer those many questions that you sent in, but again we will forward them to him for individual response. Seeing that there are no other questions, we are going to end the session a little bit early. We want to thank all those who participated today and a special thanks to the presenters for very excellent presentations. It was most interesting. Symma Finn: Before we close I’d like to make a few announcements. If you have tools and resources that you would like to share regarding communities, researchers, healthcare, public health please consider sharing them with the partners and environmental public health community by submitting them to the PEPH Resource Center. Please e-mail to [email protected] for more information. You can keep in touch with the PEPH by signing up for the listserv. And there will be several upcoming webinars, and for more information regarding the upcoming webinars and links to registration on the PEPH events page, they will be available as soon as we have finalized the schedule. The PEPH events page is at go.usa.gov/iit. After today’s seminar please take a moment to fill out the short webinar evaluation form. Your feedback is vital to helping us ensure that we are providing the highest quality speakers and information to meet your needs. And, finally, e-news announcement, sign up by e-mailing [email protected] Thanks, once again, to our presenters, Dr. van Geen, Dr. Chillrud, Ms. Golden, and Ms. Chai-Onn. That concludes today’s webinar. Thank you for participating.