Dr. Scott Evans is a tenured Professor of Epidemiology and Biostatistics and the Director of the George Washington Biostatistics Center. He is the author of more than 100 peer-reviewed publications and three textbooks on clinical trials including Fundamentals for New Clinical Trialists. His other positions include the Director of the Statistical and Data Management Center (SDMC) for the Antibacterial Resistance Leadership Group (ARLG), a collaborative clinical research network that prioritizes, designs, and executes clinical research to reduce the public health threat of antibacterial resistance as well as the Editor-in-Chief of CHANCE and Statistical Communications in Infectious Diseases (SCID) magazines.
+ Full Transcript
Rosemary Pennington: Media are replete with stories of the next big bug. Books like the Stand and films like Contagion have relied on the fear of and fascination with deadly disease to draw audiences to them, while news outlets regularly feature stories about the latest strain of drug-resistant bacteria. Researching and tracking the next big bug is the focus of this episode of Stats & Stories, where we explore the statistics behind the stories and the stories behind the statistics. I’m Rosemary Pennington. Stats & 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 are our regular panelists John Bailer, Chair of Miami Statistics Department, and Richard Campbell, former Chair of Media, Journalism and Film. Our guest today is Scott Evans, professor of epidemiology and Biostatistics at the Milken Institute School of Public Health, and George Washington University. He’s also the Director of GWs Biostatistics Center. Among his research interests are the prevention of the public health threat of anti-bacterial resistance. Scott thank you so much for being here today.
Scott Evans: Well thank you Rosemary, pleasure to be here.
Pennington: Just to get us rolling, how did antibacterial resistant drugs become your research focus?
Evans: Well, when I first started my career, I was doing a number of – I was working in HIV, doing clinical trials, evaluating interventions to treat HIV, and as time went along I saw a number of sorts of, documentaries on television and so forth that were describing superbugs, and superbug infections, and I quickly became interested in it. And back in 2011 the NIH clinical center actually experienced an outbreak of a super bug infection that infected eighteen patients. Eleven of those patients died. And as a result, the National Institute of Allergy and Infectious diseases initiated a new clinical trial- a clinical research network, called the Antibacterial Resistance Leadership Group. And I became a part of that group, as I had some clinical colleagues competing for this grant and I joined that team and we were fortunate enough to get that grant. And so, we’ve been working on this superbug infection problem since 2013, and that’s how I got started with it.
John Bailer: So, just take us back one step. How do you define a superbug? How do you know how many there are? Or what the prevalence of this is? I read in one of your papers I read that you reported estimates of like 25%-75% antibiotic use being unwarranted in acute care hospitals. So, you now, how do we know such percentages? And what are these things?
Evans: Yeah so, the general term “superbug” generally refers to a bacterium that has become resistant to antibiotics. Now, we’ve had antibiotic drugs around since the mid-1900s and the idea behind antibiotics is they are able to kill bacteria and antibiotics are very powerful and a part of our medical armament, you know antibiotics really allow us to do many of the medical procedures that we do on a regular basis. You need antibiotics if you’re going to do chemotherapy for cancer, or if you’re going to do dialysis for renal failure. If you’re going to have neonate care or intensive care unit care, and various surgeries such as organ transplantation. Any time you do procedures like that there’s a risk of infection and what antibiotics do is either treat or prevent those infections. So, the problem with what happens- what we have with antibiotic resistance is in many ways sort of a bacterial Darwinism. The bacterial- the bacteria that are resistant to antibiotics or develop a resistance to antibiotics, those are the bacteria that live to reproduce and spread. And so, we’re constantly in a cycle of bacterial Darwinism and we come up with antibiotics that can fight those bacteria, but those bacteria are smart and eventually learn to become resistant to antibiotics. One of the challenges were having right now is that, although we have many many antibiotics that have been approved by regulators for use for many many years, those antibiotics are- because they’re in common use and the bacteria are figuring out how to become resistant to them , what you really need is sort of a continual pipeline of antibiotics, of new antibiotics that can fight these bacteria. The pipeline for antibiotic development has slowed and there are many reasons for that but, now bacteria are-there are several different types of bacteria. Some of the common ones you may have heard of, Strep, what you may hear about Strep throat is it’s a bacteria staph infection bacterium, gonorrhea is a bacteria, but there are some very important bacteria that in particular have developed that are particularly difficult to treat with today’s antibiotics. Things like pseudomonas in CRE, and what happens is you can get an infection that is due to one of these bacteria, and those infections can occur in many different places in the body you can get a lung infection you can get a skin infection, you can get a urinary tract infection, an intestinal tract infection and so forth. And the hope is that we can use antibiotics to treat those infections, but as resistance to these antibiotics develop, those types of infections become more and more difficult to treat. And so, we’re looking for new ways to, or looking to develop new antibiotics or new ideas to fight these bacteria that have become resistant. A couple of the… well, one particularly interesting new idea, and it’s in the very early phases of development, is antibiotics are generally a drug; they’re a chemical compound that is made up by humans, so we start to put chemicals together and that’s how we make antibiotics. But a new strategy that were developing, is there are sort of natural enemies of these bacteria, viruses and other things, and the idea is, “can we actually use the virus?”, a natural enemy of the bacteria, to kill the bacteria. And so, we actually fight one infection, a bacterial infection with another infection by introducing a virus that attacks and eats that bacteria. And there has been at least anecdotal evidence of some success with this strategy but we need to study it in a more systematic way. But there have been a few people who have developed very serious infections which we’ve tried essentially all the antibiotics we have in order to treat those patients they weren’t able to respond to those in a positive way, and so we went to – they tried this called phages. Bacterial phages viruses that kill this bacterium and so they tried phage therapy and there’s been a number of cases where people have recovered using that phage therapy, so…
Bailer: Can you talk about the scope of the problem, I think I heard in an interview, or maybe it was one of your articles that this effects- this antibacterial resistance effects about 2 million people each year in the U.S. and that there are 23,000 deaths-
Evans: Yes-
Bailer: And how are those numbers compiled? Where do we get those numbers? How do we know that?
Evans: Yeah, so the Centers for Disease Control and Prevention at the CDC back in 2013 released a report that stated that at least 2 million people acquire bacterial infections that are resistant to antibiotics that were designed to treat those infections, with about 23,000 dying as a result. That’s in the United States. Now in the European Union there’s about 33,000 new data suggested about 33,000 people die from these infections annually. Now, the CDC is due to come out with a new report this year which will update those numbers. It’ll be interesting to see how those numbers have evolved over the past uh 6 years or so. And one of the big causes, as you mentioned earlier, one of the concerns we have is that there’s a lot of overuse or misuse of antibiotics, which only adds to the problem. So, a lot of times if a patient comes in to the emergency room with flu like symptoms and maybe they have the flu, well, the flu is a virus. And antibiotics are meant to treat bacterial infections not viral infections. And so, if antibiotics are prescribed to treat the flu, first of all the antibiotics aren’t likely to have any positive effects, and they may even have some toxicities, and they potentially promote the development of resistance to these antibiotics. And so, there’s a concern that there’s a lot of over prescribing of antibiotics, that – now part of that is that we don’t always know the – if somebody comes in with an infection like symptoms, we don’t always know immediately what the cause of that infection is, in terms of whether it’s a virus or a bacteria or something else. Sometimes we can run a test to try to figure out what might be the cause of that infection, however, the results of that test can take 3 days to obtain. There are cases where we don’t have that much time to wait around for three days for those results to come back, so we often times doctor’s treat immediately and not wanting to wait for those test results, and there are both pros and cons to that. Obviously if they have a bacterial infection you get faster treatment if you can treat immediately. However, if they don’t have a bacterial infection then you’re potentially promoting resistance and you’re treating someone with a therapy that’s not going to help them, ad could indeed harm them, you know give them a harmful side effect potentially. So, another area of research in this domain is the development of diagnostics that could get us information about whether this is a bacterium, whether this is a virus, much much quicker. And it only tells us what bacteria or what virus it is but can also test for how responsive that particular infection would be to specific drugs that we might try. And so, what we’re hoping to do, and what we’re indeed testing out are diagnostics that may be able to get us this information perhaps within a couple hours instead of 3 days.
Pennington: You’re listening to Stats & Stories. And today we’re talking public health and superbugs with George Washington University’s Scott Evans.
Bailer: So, Scott, just let me follow up on this idea that the diagnostics- so when you’re developing these procedures or evaluating these procedures, how do you know they’re aby good? I mean there are lots of different criteria in which you’d look at that. So, can you just give a quick summary of a couple of the criteria you use for say, a diagnostic procedure that works pretty well?
Evans: Yeah, sure. So, in general for diagnostics, some of the basic evaluations would consist of estimating things such as sensitivity and specificity, and- Bailer: Can you just clarify for a general listener? What do you mean by sensitivity? Or what do you mean by specificity?
Evans: So, sensitivity is the probability of a test being positive for a particular infection when indeed the person is truly infected with that disease. So, it has that disease. So, it’s the probability of the test being positive, when you’re disease positive. Specificity is the probability of the test being negative when the person is disease negative. And so, you would like both sensitivity and specificity to be very high, so that means sensitivity being very high would mean that- if a patient was walking into the doctor’s office, they want to know that “if I’m truly infected, what’s the probability the test will actually identify me as infected?”, and that’s sensitivity. Or if I’m truly not infected what’s the probability the test will show that I’m not infected, and that’s specificity, and would like those to be very very high- both of those to be very high. Now on the flip side, so that’s what I’m interested in when I walk in to the doctor’s office. Now, suppose that I have the test and the test comes back and indicates that I’m positive for a particular disease. In that particular case now, what I would like to know, is now since the test is positive, I would like to know the probability that I’m truly diseased. And so, this is what I’m interested in if I’m walking out of the doctor’s office, as you might imagine, and this is called positive predictive value. So positive predictive value is the probability that I’m truly diseased when -now conditional upon the fact that I’ve had a positive test. Where the negative predictability is the probability that I’m truly non-diseased, conditional upon, now that I’ve had a test that indicates I’m negative. And in order to calculate both positive and negative predictive value, I need to have an idea about the sensitivity and specificity of the test, and I also need to know the prevalence of the disease. And through surveillance mechanisms we may have some idea of what the prevalence is, in which case I could use Bay’s theorem to calculate the positive and negative predictive value of the test.
Bailer: That’s helpful, thank you.
Richard Campbell: I’m imagining some of the challenges that face you in your work, maybe one of them would be, how do you explain your work to doctors that don’t understand statistics in this work? And the other thing would be to journalists- so are there challenges when someone who is not a doctor is not sort of trained the way you are?
Evans: Yes, and this is really an important aspect of our work, I believe. And one of the biggest challenges I think for statisticians is how to communicate effectively with in my case, my clinical colleagues, my doctor colleagues in particular. So, the first element of that is not only getting them to understand me but for me to understand them. They help me understand the medical problem at a deeper level, and what would be useful for them as they think about how to treat patients, and how to diagnose patients. And that helps provide me a foundation or understanding of what the critical needs are for doctors who are trying to treat and diagnose patients. So, one key to that is that as I’m thinking about statistical concepts, and they’re thinking about medical concepts, we must find a common language and they can’t speak too technically for me, and I can’t speak too technically for them. We want to try to use general language for each other. However, it also requires both on my part and on their part – a little bit of learning about- so for me, I have to learn a little bit about the medicine. I have to learn about antibiotics, and I have to learn about the different bacteria and different types of infections that people can get, so that I can understand the problem at a deep level, and so that I can communicate with doctors about what I’m thinking ad how I might describe how we might design or analyze a study, or how we interpret the results of a study. So, this is a really important issue, and, I think one of the great challenges of our profession. So even though I’m not in a formal classroom, for example, in some ways, as a statistician I’m always teaching, or trying to teach people about statistical concepts so that they can understand the results of various studies and communicate with me to help me understand what they’re thinking. So, this example of describing sensitivity and specificity and positive and negative predictive values is a good example of that. So, it’s very important.
Pennington: In reading up on your career I came across the piece by D.W. announcing your taking over as director of the biostatistics center, and towards the end of that piece it mentioned that you felt like there needed to be more inter-disciplinary work in this field, and I guess coming off of Richard’s question given the work that you do, when you think about interdisciplinary linkages and work that can broaden our scope of understanding, what are you thinking about in relation to this?
Evans: So, I think particularly when we think about the superbug infection issue, it is a multi-disciplinary effort. Obviously, I interact with a lot of medical colleagues who understand these diseases as they think about treating or diagnosing patients. we have laboratory colleagues for example, who collect specimens and they run various laboratory tests. We talked about diagnostics, for example, so one of the strategies for evaluating diagnostics is we evaluate the genetic profile of the bacteria that we find. And if certain genetic traits are expressed then what that tends to mean is that that bacteria is going to be resistant to certain types of antibiotics because of the mechanism in which that particular antibiotic works. So, one of the ways we think about advancing the field of diagnostics is to get a better understanding of the genetic profile of various bacteria. And so, we start interacting with a laboratory colleagues, people working on big data problems and looking at the genetic s and genomics of both the bacteria and also human genetics, and human responses we night see, and how we can use that information to think about new strategies for how to treat or diagnose disease.
Bailer: I was going to say, I’ve really appreciated the light touch that you have in terms of describing like, randomizing Simon & Garfunkel to treatments as one of your papers. I’m sure you are working really hard to get smart bed frame, door, radar … I suspect that takes as much work as maybe executing some of these studies, but what I wanted to ask you about, was this idea of drug comparison versus therapeutic strategy comparison- I thought that was a really interesting idea, and can you give a quick overview or summary of that?
Evans: Sure, this is a – we had a recently published paper, actually the print version is just coming out over the next week or so and so as I mentioned when describing the treatment of patients- if a patient comes into the hospital or maybe they’re in the ICU, this really sort of two- well, let me back up a moment. When you’re treating patients for a particular disease, whatever affliction they might have, the decisions about how to treat that patient are – it’s not a single decision necessarily. In many ways the treatment of patients is dynamic- you make an initial decision about how to treat a patient but then you see how the patient is doing. They may be responding well, they may be responding poorly, you get more information about how the patient is doing and then with that information you might make an adjustment to treatment. So, what happens with patients with infection, it’s really sort of two very well-defined treatment decision types. The first is what infectious disease clinicians called empiric therapy. That this is the time which I haven’t been able to run these lab tests yet, so I don’t really know what the bacteria is or whether it’s a bacteria or a virus or if it is a bacteria what type of bacteria it is. And I don’t know the profile about which drug I can treat it with. But because some patients can’t wait you have to make an initial treatment decision about how to treat them. And one option for that is maybe I don’t treat them at all, maybe I wait for specific information to come. But I may take a lab specimen and send it off to the lab, but I may have to wait 3 days for that lab information to come back. So, I make an empiric treatment decision when I first see the patient. Then about three days later I get back the laboratory information, and that laboratory information can generally tell me what the pathogen- what the offending pathogen is. What bacteria or virus is responsible for this infection and may also give me information about which drugs might be effective at treating that patient. So instead of studying – and that second decision type they often refer to this as sort of definitive therapy. Your – so we have this two decision time points. Now, typically in clinical trials and most commonly in clinical trials anyway, we might randomize patients to one of two or more options. and then evaluate how effective and safe those options are by following the patients that are randomized into those options and evaluating them. But in this particular case, because there’s these two time points that are really critical for decision making what we’ve done is basically said, “well, let’s study strategies of treatments rather than specific treatments”. And what I mean by that is, suppose I come in and a doctor doesn’t know whether I would be susceptible or resistant to a particular drug, but he has to treat me now. Three days later we’ll get some more information, but he has to make a decision now. So unfortunately, we’re stuck in this situation where you sort of treat today and you diagnose tomorrow. So, they start me out on a particular drug, hoping that might be the right drug. Three days later we get back information that tells me from the lab whether they believe that was a good choice or not. Whether they believe I’ll be susceptible to this drug or if my infection will be resistant to this drug. And if I’m susceptible then maybe I stay on that drug and that looks like a good path to follow and I would go that route. However, if the lab comes back and says “well, it looks like your infection has resistance to this drug”, then they would make adjustments, and say, “well listen, we have to change you to something else, because it looks like you’re resistant”. Now, what that means is that I’m studying a strategy of starting out with that drug, but the strategy allows for an adjustment if I get new information that suggests that an adjustment would be wise. And so instead of studying just one drug at a time, I study strategies of treatment that allow for adjustments as new information arises, such as new laboratory information that tells me whether I might be resistant or susceptible. Or even just clinical progress if I seem to be progressing in the right direction, then maybe you keep me on that therapy. If I’m getting worse then you make a change and make an adjustment.
Pennington: Well, Scott, that’s all the time we have for this episode of Stats & Stories. Thank you so much for being here!
Evans: Well, thank you it was my pleasure.
Pennington: Stats & 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 or apple podcasts or other places where you find podcasts. If you’d like to share your thoughts on the program send your email to statsandstories@miamioh.edu or statsandstories.net. be sure to listen to future editions of Stats & Stories, where we discuss the statistics behind the stories and the stories behind the statistics.