Lance A. Waller, Ph.D. is Rollins Professor and Chair of the Department of Biostatistics and Bioinformatics, Rollins School of Public Health, Emory University. He is a member of the National Academy of Science Board on Mathematical Sciences and Analytics and has served on National Academies Committees on applied and theoretical statistics, cancer near nuclear facilities, geographic assessments of exposures to Agent Orange, and standoff explosive technologies.
+ Full Transcript
(Background music plays)
Pennington: Public health researchers have a lot of information to manage. Beyond simply tracking how health issues spread, there's environmental information to gather and decipher as well as socio economic or infrastructural data to analyze. The mapping of such information can help researchers and medical professionals better manage a public health issue. That's the focus of this episode of stats and stories where we explore the statistics behind the stories and the stories behind statistics. I’m Rosemary Pennington. Stats and stories is a production of Miami University's departments of statistics and media journalism and film as well as the American Statistical Association. Joining me in the studio is regular panelist John Bailer. Richard Campbell is away today. Our guest is Lance Waller. Waller is Rowland's professor and chair of the department of biostatistics and bioinformatics in the Roland School of Public Health at Emory University. Thank you so much for being here today.
Waller: A pleasure to be with you.
Pennington: In a video I was watching earlier I heard you sort of describe where your work is situated as the field of disease ecology. So before we start talking about mapping or biostatistics could you just sort of explain what disease ecology is?
Waller: Sure I'm happy to do so. When we talk about health and disease, usually talking about a combination of some pathogen. Well, that's a virus or a bacteria and a host which we're most concerned about human beings and ourselves, so if we get infected we become sick. And ecology looks at the combination of how living and non-living things interact together in an environment situation. So, for a disease to transmit from one person to another, if we have a cold, we might cough, the pathogen is in the air and touched or breathed in by somebody else. Disease ecology is trying to understand the way that works. More complicated systems like malaria you might have a person who is infected, get bitten by a mosquito which then becomes infected and then that pathogen has to go through part of its life cycle in the mosquito and it's transmitted to another person. So you can imagine mosquitoes need a certain environment to live, people need to work in certain areas where mosquitoes might be and all of this has to line up at the right time and right place for transmission to occur and the field of disease ecology can focus on how all those different systems interact.
Bailer: So could you talk a little bit about how disease ecology relates to the idea of things like drug abuse and the opioid crisis that we've been talking about and I know you've done some work on the spread of methamphetamine abuse. So how does that fit into this model that you just described?
Waller: In some ways, what's missing is the actual pathogen. But we're still talking about a problem that requires different individuals to interact with each other and an environment within which that interaction can happen. So in the term of illegal drugs or you have a new drug problem occurring in a new population you have people who are susceptible to use or abuse the drugs so. In one case opioid and methamphetamines are another but you also have an environment within which people have become addicted to say the painkiller they're off of the coverage they might be under their insurance but they're still suffering from pain and looking for some relief for this. This might put them in a situation where they would look for some other outlet for obtaining the drug, a less regulated outcome that can lead to less regulated use and overdoses and so on. So it's an environment, which can be the people you know, it can be the insurance background we are in, so you can think broadly about the social environment as well as the physical environment and you can think about the transfer of the problem you're interested in from one place to another. So in the spread of methamphetamines we were looking at a pattern where the drug was relatively cheap to make in small batches and so this started more in rural areas then moved into an urban area rather than being a big expensive importing product that started in urban area and then spread out to rural areas.
Bailer: So when you think about some of the…you talk about susceptibility and the transfer of this particular concern. What…are you interested in modeling some of the factors that might be associated with such susceptibility?
Waller: Yeah, that's a big part of what we're interested in. On the epidemiology side there might be risk factors. For instance for an individual, what makes a person more susceptible to either…if you think about an individual who had been taking prescription painkillers for a while what's what is the situation that would put them in a position for wanting to reach out and purchase something illegally. Similarly you know, what is the situation around you? So you can think of risk factors as being you personally but also the environment you're in. I might not know where to purchase illegal opioids today but do I know a person I could talk to who would know a person who could talk to you know a person you could talk to and depending on how desperate I would get, you can think about those sorts of things. So susceptibility is based on your own personal inclinations, your own personal habits as well as what's available to you in the environment around you.
Pennington: So I understand how you could use mapping and public health to track you know where something is spreading or maybe where something started, but how do you map something like susceptibility? How do you sort of forecast where something may spread and map that out and sort of track it geographically?
Waller: So there's a couple of ways to do that. One is through all research is learning from you know standing on the shoulders of giants, learning what people have done before. So you may have studies where some individuals with a problem have been intensively interviewed and you look for the common features that go together. This is very standard epidemiologic. You know, what did people have in common? Oh they always ate chicken salad at the church picnic as a standard epidemiology kind of disease detective problem, but it's very similar here. You might look at people where you can get a lot of detailed information, start to narrow down what the risk factors were, what common features they had, that the cases had, the people who did not get the disease did not have, and then you can say where are those risk factors most prevalent in the United States. So this can be done either through looking for census data or describing the socioeconomic background of it. Something you have more national scale data or state wide data on or a county wide data. It could also be through mathematical models. So there's a lot of differential equation mathematical models of the spread of infectious disease and some of these go back to the…the basic structure goes back to the early 1900s in modeling malaria. But there's still a lot of mathematical modeling work that is done so you're simulating outbreak. If you are simulating the outbreak if you can test different response mechanisms, so where should we be distributing the vaccine, how should we do that? There was a lot of work on how best to distribute the Ebola vaccine in 2014 in the outbreak that was in the news a lot and the World Health Organization had several groups of people working with some really top notch mathematical modelers. Other people have looked at the design of vaccine distribution in terms of responding to an outbreak of something like Ebola.
Bailer: So can you talk a little about what does it mean when you say simulating outbreaks?
Waller: It's like a big computer simulation game. It's like a Sims game which I guess probably dates myself.
(Collective laughter)
Waller: But if you think about a Fortnite to make it more example…you may be controlling one character, but in the computer game as it were, there are characters interacting, things are happening around you. There was actually a very famous World of Warcraft outbreak that they had simulated disease and they had set the parameters for how fast it would spread and how deadly it was, in such a way that it wiped out a bunch of people's characters!
(Collective laughter)
Waller: Games, the way people played the game is they got information that voided a certain part of the imaginary world. Now you can do that by setting parameters that are very close to say how influenza spreads or something like that and you can see how little changes in a complicated system may have big effects.
Pennington: I heard somebody talk about something really interesting. The difference between information and data. And I was wondering if you could talk a little bit about what you see those differences being and why understanding the differences between those things is important for your field?
Waller: So we're living in a very data rich environment now. There are a lot of people measuring a lot of things or you can search on the Internet and find some data about the project you're interested in. You want to look for flu information and flu data, you may find counts of data that were reported every week. That comes out from the Centers for Disease Control. Trying to understand how fast it spread, you might want some more detailed information that a dataset may not have. So one example of this is if we look at the number of cases of infectious disease that are reported every week. So you get one case and then there's five cases and there's fifteen, then there's twenty and then there's fifteen and it goes down, that tells you how many cases happened. But if you want to know, what are the factors that make it more or less likely to transmit between people, should we say wash your hands more, should you wear a mask, those kind of things, you need to know how many contacts are made and what's the probability of transmission per contact? If you just see the number of cases that happened you can't tell the information of whether it's very infectious and there are only a few contacts or it's not very infectious but the people who got it had lots of contact. Those two extremes may generate the exact same number of cases. So the data tells you the number but it doesn't give you the information…that number alone won't give you the information about the infectivity and type of transmission. So data contains information but it may not contain the information to answer exactly the question you want to ask. So in classes I often tell students and my colleagues, you start with some public health question you want to answer. How infectious is this disease and you should think what would be the perfect data to get? Well I want to know, who doesn't have the disease and who they were exposed to, how many people with the disease were they exposed to, when did they get sick and so on but then there's often the data you can get which might be the number of cases per week and it's not exactly what I want. So now I can answer a slightly different question, how fast did the number of cases break out and I should see if the question I answered is close to the question I asked…I wanted to ask. If it's not, I’d say what other data would help me kind of disentangle this puzzle?
(Background music plays)
Bailer: That's a great description! I like that a lot. I'm probably going to steal that.
Waller: Yeah well you know I grew up reading Encyclopedia Brown and Sherlock Holmes, so the idea of being a detective who spends all day in my office was really appealing.
Pennington: You're listening to stats and stories where we discuss the statistics behind the stories and the stories behind the statistics. I'm Rosemary Pennington with Miami University statistics department chair John Bailer. Today we're talking about the mapping of public health issues with Emory University's Lance Waller. Lance, there is a story that I've heard from other public health officials I used to do some reporting on health and science stuff, when I worked in Birmingham Alabama and I was talking to someone who actually uses remote sensing in public health issues, but she went back to this discussion of a cholera outbreak and John Snow, I know you've written about this as well. So could you just sort of I know we're talking about sort of statistics and mapping now but maybe sort of dial us back a little bit and sort of explain kind of how this idea of mapping public health issues sort of grew up and how John Snow in particular is such an important figure.
Waller: Yeah, so John Snow was an anesthesiologist living in London in the mid -1800s and London, at the time was one of the biggest cities in the world and connected through its trade routes which brought wealth and goods from a lot of places but also brought infectious disease and so London would experience cholera outbreaks every other year or so when it would get in the summer. And this is before the germ theory of disease was widely accepted. The microscopes were just being developed, the vibrio cholera, the microorganism that causes cholera have been spotted in Italy but the idea hadn’t really been around and the big competing theories were that the air was bad, and if you think about it, it's hot, it's summertime in London in the 1850s, the sewage system is developing but not perfect, so the air just smells awful. People are getting sick and the idea was truly, those are related. They would count the number of cases, they'd see it jump up during the outbreak and then fall off. The London Board of Health was concerned about this and they were wondering what they should do and one group of people thought it was due to air and there was a hypothesis put out that some of the storm drains, the sewer system under London that was developing had cracked into a plague pit, a mass grave of a quake outbreak from several hundred years before and maybe that air was seeping up through the drains in the street. And then John Snow who lived in the Soho neighborhood in London he was a physician and they were keeping track of where the deaths from cholera occurred during that outbreak and they noticed this one particular block had a very large number of cases and roughly centered around one of the public water pumps on Broad Street. And so there were several maps, Snow’s was not the first map but during the period of time that the Board of Health and the neighborhoods were looking into this, there were several updated versions of the map. Snow was a proponent that the water was causing a disease, he petitioned the Board of Health through a lot of meetings and they ultimately removed the handle of the water pump kind of as the outbreak was waning, we kind of condensed this to a story almost a public health fairy tale that, look at the maps, surely it's the pump, take the handle off and save the day and I'm not making light of Dr. Snow's work because if you read his papers, they are surprisingly similar. He's often discussed as one of the founders of modern epidemiology for his careful data collection, his analytic approaches. So it's well worth your time if you're interested in the sort of thing in reading Snow's cholera papers, but his book, which is two of his papers contain two maps, one of them is a neighborhood map that has tally marks, so the deaths that each site in the Broad Street pump is marked right there. The other one is a map of which water system one of which collected its water upstream from the majority of people in London where the sewage was being dumped in the river, and the other downstream and he looked and noticed that the rate of disease in the water pumps was the water sources was much higher in the group with the sewage contaminated water than that so that's very similar to the epidemiology studies we do today. The dot map of the tally marks, Snow’s map which shows the water pump in the middle, there is also a map by the group proposing the bad air hypothesis and that shows drains and if you think about it there's always going to be a drain next to a pump in case the pump leaks. So the maps look exactly the same but people looked at it and interpreted it very differently.
Pennington: Oh that’s really interesting.
Waller: Yeah so Snow won the day in being right and we tell that story a lot but some other the reports of the time say well clearly Dr. Snow's map shows this was caused by bad air coming out of that particular drain. So science marches on and we remember the stories and I think the thing with Snow's…you know maps are great communicators, people are always excited to see where their house is and things like that. The Snow maps really should paint a picture of that sort of stunning, in your face when it's right in front of you but I always try to tell the story about the other interpretations because we always look at these maps through our own lens and we tend to see what we want to see and part of doing public health research is to try to figure out in what ways might I be mistaken, what is the most consistent…what's the story that's most consistent with the evidence I have, and then try to back that up with follow up studies. So Snow had that one map that gets a lot of attention but the map of the water source is really kind of sealed the deal that water was involved in the transmission of cholera.
Bailer: What a great retelling of that story Lance, thanks! That's a lot of fun. As someone who's had a chance to visit Broad Street and the John Snow pub and then…
Waller: You should point out the map’s still useful because you know John Snow pub is at the location…
Bailer: It’s right…this close to the pump I mean…
Waller: It's now called Broadway street but it is…there's a book there you can sign that epidemiologists around the world like to put their name in.
Bailer: And you can actually join the John Snow society there.
Waller: Yeah.
Pennington: It's great!
Bailer: Its good fun. But I have to ask you, what is your favorite map, Lance?
Waller: My favorite map, well I like John Snow map. And then I'm kind of a map nerd so.
(Collective laughter)
Waller: In my office there's always an antique map catalogue out every year and it's always got good ones and that's have…every one of them has something, and when I became department chair nine years ago and I just stepped down the end of August I went to a map gallery and they took some pictures to talk about it and I found a map there that is from an atlas from the 1800s. It was in the bargain bin so the frame is more expensive than the map. It's kind of an odd combination of things. It's got North and South America with where the mammals live, has an inset next the south America of what plants grow at what elevation, there's an inset of Austria for a reason I can't…and the third that shows Java and Sumatra and there's a line drawn on there and in German to one side it says Asian animals or Asian fauna and to the east it says Australian fauna which is something called the Wallace Line. Wallace was a contemporary of Darwin's who noticed that with continental drift the marsupials go part of the way towards Asia but don't get all the way to Asia similarly apes and apes monkeys from Asia don't get all the way to Australia and there's kind of a wind through the archipelago where this strange happens. So I've always heard it in ecology classes as the Wallace Line and then I notice the date on the map is about ten years earlier or twenty years earlier than Wallace's paper.
Pennington: Oh wow!
Waller: I’ve kind of stumbled into something that I thought was really interesting of course there are some history of science papers written about you know older versions of Wallace's Line and there's this map shows up in it, so that's really fun.
Bailer: That is cool.
Pennington: You raised, when you were talking about John Snow the way the news covered his work. I wonder if you could talk briefly about sort of how you think journalists do now when they're covering public health issues.
Waller: Yeah I think there's a lot of health journalism, there's a lot of public interest. I think there are so many studies that a lot of people are kind of dismissive of a sentence or article that starts with a study shows because people say that all the time. On the maps side, once you see a map, you tend to link it to your own personal experience of where's my house syndrome. And you know most major news agencies will have a staff of geographic information systems specialists who put the maps together. The New York Times has a lot of you know they have a great graphics department and they do a lot of interesting maps, as we come up to the midterm elections there's a lot of mapping of voting and trends and who's voting for what and where do they live and how does it look where you live so you'll see more and more of those. So I think the map is a great communicator. It's not error free because you know a really good map of bad data is a lot more believable than a bad map of good data.
Pennington: That’s a good point.
Waller: And we can make really great maps, so you know I think just like any report you see, it's good to have a healthy skepticism and try to dig a little deeper, see what data are behind that map. Am I seeing different things at local levels than I see at bigger levels, so I think there's a big public interest you know people are always interested in what improvements can they make, they'd like to be healthy nobody wants to be sick how can I get the care I'm looking for. Why is you know there’s a lot of interest now why is the opiate opioid epidemic bad why is it bad here and not there. What are the local factors that may drive that, how can we make a difference in our own community. So. I think people want answers to these questions and I think it's important when we report these, to try to dig through some of the science jargon that we tend to use when we're talking to other scientists and try to put that in plain but accurate language.
Bailer: So what are some of the most common mistakes that you see in maps, the mistakes that really drive you crazy?
Waller: Well some of the election maps. So they color them red and blue and we've all seen these. You can get overwhelmed by one color or the other based on the size of the county western U.S. has very geographically large counties and in the eastern U.S. all the counties are about the same size. So we tend to take the overall color into account without adjusting for how many people live in certain places. When AT&T and Sprint started out with some of their coverage in the US, AT&T covered more people but Sprint covered more area and there was an ad campaign where they showed the two maps and show that the Sprint color or Verizon color I forget which one it was covered much more of the map so it surely got to more people, but AT&T tried to say it was they covered more people but the visual impact of this was just you know you couldn't overcome the sort of in your face picture of it.
Pennington: So what advice would you give to journalists who are trying to cover public health stories and sifting through data, you know, besides sort of just you know digging deeper for the data behind maps, if there are maps included in the studies they're covering, what other things you think journalists kind of need to keep in mind as they're reporting on stories about the opioid epidemic or you know, the spread of something like you know Ebola.
Waller: I think an important thing would be just you know journalists often have a collection of sources and resources and people they talk to. School of Public Health and Public Health researchers, some of us are easier to talk to than others.
Pennington: Yes you are, that’s very true.
Waller: It's much better when they're putting complicated ideas into straightforward and accurate language I think you know one of the keys to being good at any field is who are your other experts, who are the people you can talk to. So I think having a contact who's familiar with epidemiology is really important because epidemiologists work with chronic diseases like cancer, they work with infectious diseases and global spread of diseases that you know maybe more rare but frightening and trying to put those things in perspective. So I think having some good sources to go to even for just offsetting kind of how you can make a terrifying headline but is it really a concern to an individual I think most people are reading these thinking about what does this mean for me and my family and my friends or what does this mean in terms of policy if we change this policy and this is going to happen if we add something else could we alleviate a problem. There's a lot of discussion on that and as a journalist I'm sure you hear things on both sides. It is interesting as in the middle where you can try to figure out which part of each person's story is accurate and reliable. So I think having good people who can dig into the details and know what kinds of mistakes could be made and how to avoid those is important.
Bailer: So if someone wants to do what you do for a living, what are the things that they should consider studying, how can they prepare for the type of work that you do?
Waller: Well I think statistics is a good field if you have broad interests because everybody has a data and wants to do something with.
Bailer: Amen, brother.
Pennington: I was like good one, John!
Waller: If you're interested in health issues, whether those are social determinants of disease or infectious disease, Schools of Public health provide good training and introduction to epidemiology. I minored in epidemiology in a veterinary school so I was exposed to a different set of diseases, but I think that’s gotten me more interested in the ecology aspect than if I had been in a large medical center taking classes. So Schools of public health often have training programs, the National Institutes of Health, Center for Disease Control have a lot of interesting information that can get people started, background reading on epidemiology and public health and disease detectives, I think kind of reading some of the descriptions of these can be…it can get people interested and then in order to work with the data you know experience with computers and experience with math and statistics are going to be essential for trying to make good decisions and understand what stories the data have to tell and try to get information out of data.
Pennington: Well that’s all the time we have for this episode of stats and stories. Lance, thank you so much for being here today.
Bailer: Yeah! Thanks Lance.
Waller: Thank you! I enjoyed it.
Pennington: Stats and stories is a partnership between Miami University’s departments of statistics and media journalism and film and the American Statistical Association. You can follow us on Twitter, Apple podcast or other places where you can find podcasts. If you’d like to share your thoughts on the program, send your e-mail to statsandstories@miamioh.edu or check us out at Stats and Stories dot net and be sure to listen for future editions of Stats and stories where we discuss the statistics behind the stories and the stories behind the statistics.
(Background music plays)