The Role of a Biostatistician in a University Setting

 More information related to this Podcast


Guest:  Dr. Andrew Lawson – Biostatistics & Epidemiology

Host:  Dr. Linda Austin – Psychiatry


Dr. Linda Austin:  Dr. Andrew Lawson is Professor in the Division of Biostatistics and Epidemiology, here at MUSC, with an interest in using statistics to determine the geographical distribution of diseases.  Dr. Lawson, in this podcast, maybe you could explain for us just what the role of a biostatistician is in a university setting.  How is it that you help researchers do what they do?


Dr. Andrew Lawson:  We have a very wide-range role at the university in general.  We help out many researchers with different projects, in different settings, both in the clinic and bench scientists who work in laboratories, and so on.  And, also, the population level, we actually analyze studies where people are interested in disease in the population, and not just in the hospital, but outside, in residential locations. 


So, we help in a lot of different ways.  One is in clinical studies, where we’re looking at the progression of disease over time, and looking at the effect of drugs, and so on, or treatments in general.  And we actually handle the number crunching, if you will.  We actually take the numbers and look at them, and decide whether there’s any significant effect of a drug, or a significant benefit of a particular treatment in the clinical setting. 


Also of interest is whether there are big differences between different parts of the population.  For instance, in South Carolina, you might find that down in the Lowcountry, there are a lot of people with one particular disease, and in the Upstate, they have different diseases.  So, why is that?  What is the reason for that?  And, also, do we provide services for these different diseases in different areas?  We use numbers.  We essentially take the numbers out of these problems and say, can we actually make some inference?  Can we make decisions based on collecting these numbers and processing them in terms of, say, health service resource allocation, that is, providing funding and resources for services for particular diseases, or for simply finding out new aspects of a disease itself, i.e. why are there more cases in Columbia of disease X than disease Y?


Dr. Linda Austin:  Let me take a very specific example.  I have friends who live in a very small community in Maine.  They became concerned because, in this very small community of 200 people, there seemed to be a lot of people getting cancer.  In fact, on one street, there were five women who got different forms of cancer in a two-year period.  And they became concerned, could there be a contaminant?  They live near a facility that was causing contamination, and they began to believe that there might be a cancer cluster.  This is the sort of question, right, where you might be called in?  How would you go about answering that question for them?


Dr. Andrew Lawson:  That’s a very interesting point.  That is an area in which a lot of time is spent working on solutions to the problem.  And, basically, we often have to deal with things called cluster alarms.  What you described is what’s called a cluster alarm, which is when somebody notifies a health authority, or experts, that there is, locally, some very poor health indicator, meaning, there might be some industrial works or factory or, say, a waste dump site, which people believe to have an impact on their health.  And it could be for all kinds of reasons.  They might think it’s air pollution that’s causing it.  It might have to do with the water supply, anything like that. Various types of pollutants might get into, essentially, their living environment.  So, people are often very concerned about that, especially when they see large numbers of cases of a particular disease or something in a very small location.


What can we do about it?  Well, very often we collect data around that source to find out whether the cases form some kind of pattern.  For instance, if there is a real link with a source like that, one would expect there to be lots of cases near it, and then much fewer cases further away.  So, we look for the spatial patterning in those cases to see whether there’s any linkage or not.  But usually, also, we look at a whole lot of different diseases to find out whether, in fact, there’s a general health problem in the area, rather than just a specific health problem having to do with those diseases. 


Now, we also look at a much bigger area as well because sometimes studies that are focused on one particular point on a map can actually lead to very peculiar and erroneous conclusions, so we have to be very careful about how the data is analyzed.  Basically, we collect data simply by looking at the cancer registries held by departments of health.  We look at how many cases are admitted to ERs, in case we’re looking at, say, respiratory complaints related to air pollution.  And we count the numbers that are close to and far away, or count the numbers that are close to a particular wind direction, another thing that we might be interested in with air pollution.  And once we do that, we carry out some tests to find out whether these are really, in fact, significant, or whether they’re not significant at all. 


Dr. Linda Austin:  So, I guess, maybe, one way of thinking about it would be, let’s imagine that you went up in an airplane and you dropped a bag that had a million pebbles.  You could predict that those pebbles would not be perfectly distributed, that some areas would have more than others.  Just by chance, that could happen.  So, your job, as a statistician, then, would be to figure, well, was it just by chance, or were they related in some way?  But it gets to the heart of the problem, that things that happen just by chance are not perfectly, evenly distributed.  So it’s a question of when is it that versus when is there actually some reason for things to be clumped together.


Dr. Andrew Lawson: Yeah.  I mean, basically, we have to sort of tease out the difference between those two things.  If there was no problem with the pollution locally, we would expect there to be some kind of random distribution of the disease.  It would be randomly varying.  And there are some diseases that are known to actually cluster without there being particular sources for them.  For example, a very difficult disease to deal with is childhood leukemia.  That disease is known to cluster in different parts of towns, or different parts of counties, without there being, necessarily, a source or an environmental hazard that might be related to it.


Dr. Linda Austin:  Now, why is that?


Dr. Andrew Lawson:  As such, right now, there is no answer to that question.  There is still no definitive statement, or solution, to what the etiology of childhood leukemia is.  Some people believe it might be viral.  Other people believe that one of the causes might be some stressors in the environment for children who have, for example, moved into an area very recently.  One of the theories now is that leukemia risks are to do with people who move.  Actually, mobile families, often, are much more at risk of having children getting childhood leukemia than families who don’t.  So, there are a number of possible explanations currently available, but there’s no definitive answer.


Dr. Linda Austin:  So, stress in general might cause leukemia?  So, anything that stresses a population locally might be a cause?


Dr. Andrew Lawson:  It might be a cause.  But, in studies where they’ve looked at things like nuclear reprocessing plants, when they’ve looked at these plants, they’ve found that people who moved in to work at the plants were much more at risk of having children who got childhood leukemia than local people.  It appears to be something to do with movement and the moving in of these people rather than being there.


Dr. Linda Austin:  Could it ever work in the opposite direction, that there might be something about an environment that is protective to a population, and then when you have newcomers moving in who don’t have the benefit of that protection, they’re at higher risk for a disease?


Dr. Andrew Lawson:  Yes.  That could also be a factor.  In fact, it might be a factor in this case, that the local population, in some sense, is being protected, and newcomers are at risk.  Absolutely, yes, that could be the case.  But, one of the difficult things about this is if you, in addition to knowing about the childhood leukemia, for instance, if you had to actually find out whether a pollution source actually added to the childhood leukemia, you’ve already got a problem because the existing population may already have clusters of disease, and this was just adding to those clusters.  It makes the problem doubly difficult to actually tease out.


Dr. Linda Austin:  Well, you certainly have a fascinating area of specialty.  And it sounds like you really get to learn about so many different areas of health and medicine.


Dr. Andrew Lawson:  Yeah.  Oh, it’s certainly very interesting, and it’s one that’s very applied.  I mean, I’ve worked with lots of different health departments and collaborated with lots of people in these projects.


Dr. Linda Austin:  Dr. Lawson, thank you so much for talking with us today.


Dr. Andrew Lawson:  Thank you.

Close Window