This is Clifford Mitchell, and I'll be talking about approaches to Possible Clusters, Case Definitions and Exposure Assessment. When you are starting an investigation of potential cancer cluster, one of the first things you need to do is think about what is the case definition. The reason this is important is it determines everything else that happens. What is the condition you're talking about, what's the concern, and how can you be sure that what you're looking at is what you think is the cause or issue of concern? So, let's talk a little bit about what the case definitions are, about case confirmation and then, what are the considerations both from an epidemiologic and exposure assessment point of view and then thinking about things like laboratory confirmation, and then biological plausibility. So as I said, establishing a case definition is the first and most important thing that you want to do because not only is it incredibly important, but it will have a bearing on everything that you do from there on in. So, if you have a good case definition it will be possible for you to be more confident about whatever your findings are. If you don't have a good case definition, or if your case definition is ambiguous, that may still be valuable but it's important to realize that the looser the case definition is, the more ambiguity you may have when you finally get to the point where you're thinking about, "Is this something that I need to act on, or is it not something I need to act on?" So, when you're talking about a case definition, one of the first things you want to do is look at what is the basis of your case definition. Is it on the basis of a diagnosis, or is it on the basis of symptoms? If it's on the basis of a diagnosis for example cancer, one of the questions to ask is, are you talking about many cancers, are you talking about a single cancer, or a single cancer type, or even a single cancer sub-type? If it's on the basis of symptoms, are you talking about a syndrome which has very clearly defined symptoms, or about one in which you have multiple symptoms and you're trying to describe what might be considered a syndrome, or a constellation of symptoms. These are all things that from a clinical point of view are complicated. But when you're talking about a potential cluster, it's also complicated because as I said, the broader your definition, the more people who are considered a case, the more difficult it may be to figure out whether or not a case, or a syndrome, or a set of symptoms is linked to a specific exposure. So, one of the other things to think about in this regard is, "Where you sit, determines what you see." Oftentimes, when engaging with a community, or when you are yourself a member of the community, you may be looking at a number of people coming to you talking about a whole range of different symptoms rashes, headaches, cough, and it's difficult to know whether this is all the result of a single thing, or a group of things or exposures, or maybe there's flu circulating in the general population and so you're seeing a lot of things happening commonly. So, oftentimes when you are looking in such a situation, you may find that not being too restrictive in your case definition too early allows you to have a broader picture of what are the concerns of the community. So, as you are thinking about a case definition, always make sure to balance the how much are you listening, how broadly are you considering the case, and then at what point do you say, "Here's the point at which I'm going to establish a case definition and from here on in everything that meets the case definition will be considered a case. If it doesn't meet the case definition then we're going to consider it related to something else." That's why it's so important to start with listening to the people who are talking to you. The people who come in to say, "I've got headaches. I've got nausea. I've got a rash." They may not think about this in terms of a single exposure, or a single cause, or being part of a cluster. But it's very important to think about a possibility that you want to be broad in terms of your set of symptoms. At the same time once you are listening, you have to be listening with part of your ear and then with part, you have to be thinking, "Okay. Does this sound like other things I've been hearing? So, this is a complicated issue." So, why is this such an important and difficult problem? Well, because as I said, "Once you settle on a case definition, that determines pretty much whether or not you think at the end of your analysis that your findings are suggestive of some association between the case and people who meet the case, and that what you think might be the exposure, or set of exposures. " So, if your community or individuals have many definitions, that's fine. At some point or another though, you may want to do something like what's described on this screen, and it describes why it's so difficult if you have a very broad case definition to link that case definition to a specific exposure. So, if you look at the box that's on the left-hand side, this would be considered the ideal and if you had a group of people all of whom were exposed to a specific thing, and all of whom meet the case definition, and then another group of people none of whom is exposed to the thing, and none of whom meets the case definition, this is a perfect example of an association where you can say with great confidence there is some relationship here, more likely than not between exposure and whatever is my case definition. By contrast, look at the box on the right. In this box, you have some people who meet the case definition who have exposure. But you also have some people who meet the case definition who have not been exposed to the thing of interest. Similarly, at the bottom, you can see that there are people who are not necessarily meeting the case definition, but some of them are exposed and some of them are not exposed. So, there's a lot of overlap here and it's much more difficult to discern whether or not there's a true association between the exposure and the case definition, or the outcome, or the health outcome of interest. So, the cleaner the case definition, the easier it is to test whether or not there's an actual association. Say you're in a situation where somebody is saying, I think that there is an association between some health outcome or some condition or set of conditions and an exposure. In many cases, individuals with whom you speak either in the community or elsewhere, may have an idea of a potential case definition. That case definition could be broad or it could be narrow. It could be I'm concerned that cancer is associated with living in this area or I'm concerned that lung disease or rashes or something else is associated with this particular either a factory or hazardous waste site or something else. It depends very much on whether or not you can get a very good definition as to how your investigation will proceed, but as I said, first you want to listen very carefully and hear what people are telling you, and then they want to think about, so should we include multiple cases or multiple different kinds of conditions within our case definition? Now, some of where you make this decision will be based on how inclusive you want to be and how you may want to allow yourself a lot of flexibility, and that's okay to do. One of the other things that you can do though, is to think about biological plausibility. Biological plausibility means, from what we know of the relationship between a particular exposure agent, and what we know about how that biologically manifests itself in a human being or a group of human beings. Do we think that it's likely or plausible or even possible that such an exposure would cause such an outcome? There are many examples where you can talk about biological plausibility, and you can say, well, it's certainly plausible that exposure to viruses can cause a host of symptoms which include rash, cough, fever, other kinds of things. There are other kinds of biological plausibility where you could say, for example, exposure to sound waves is not likely to cause systemic disease for a variety of reasons based on what we know about biology and the mechanism of sound waves. However, here's an important thing to think about when you're thinking about biological plausibility. Biological plausibility is inherently a conservative way of thinking about the world in some respects, because it's only based on what we know of biology up until the point, where we think about, is it biologically plausible? Here's a good example. Many years ago we used to think that ulcers were caused by stress. We didn't think that there was a lot else that caused ulcers, but one very well now known pathologist, hypothesized that because he saw it in many of his autopsy specimens, there were bacteria present that could potentially cause ulcers. When he proposed this, people who were talking about biological plausibility said, "That's ridiculous. There's no way that ulcers can be caused by a biological organism. That doesn't make any sense." Well, it turns out he was right, and in fact, it's now a standard treatment to treat some kinds of ulcers with antibiotics, because they are known to be associated with these particular organisms. So, the problem with biological plausibility is that it doesn't necessarily account for the fact that you might not know the mechanism by which a disease is caused and you'd need to leave yourself a little bit of room to think that you might not know everything, and so don't automatically reject something because it doesn't fit your current state of biological knowledge. So, we've talked a little bit about case confirmation, once you do have a case definition, now you've need to find additional cases. This is an important step because the more cases you identify, the more completely you're confident that when you then say, okay are these cases in some way related to this exposure? That you've captured the universe of cases that are out there. So, one of the issues in finding cases is how easy is it to find them, this is a problem that public health people struggle with commonly. Oftentimes, there may be an existing source of administrative data. For example, there are registries for cancer, births, deaths, now more recently there are electronic health records which may allow you to access specific diagnosis like diabetes, hypertension, and other kinds of diagnoses, pneumonia's or other conditions. The challenge with these is again first, you have a good case definition. Secondly, do you have access to those electronic health records and those data? Then third, can you apply your case definition to identify the cases. These are always questions that are complex, they often require interactions with data custodians such as the Vital Statistics Administration or your electronic health records administrators. In many cases, these records may be protected or have privacy protections associated with them, so that it's important to have that administrative clearance before you can access the records. If you do have access to those records, that's very helpful particularly if your case definition involves specific diagnoses or conditions because these administrative records almost always are on the basis of specific diagnoses or specific events such as birth, death or a specific diagnosis or illness. Sometimes, however, case confirmation and case-finding are more [inaudible] of the paths if you will. Because, let's suppose that you have a case definition which is fairly broad or involves a set of symptoms rather than a specific diagnosis. In that case, there are other ways you may look at finding cases, you can talk to neighbors, so you can talk to health care providers, you may directly participate in case finding, and it's been more and more common for community advocacy groups, and community members to use social media, community mapping, and other tools to go directly to people who might be potentially affected. This can be a very valuable tool especially if you're not talking about a specific diagnosis. However, when doing this, you also have to be concerned about privacy and confidentiality concerns. Bear in mind that even if you as an individual member of the community are able to ask questions of the public and say ask for volunteers to give you specific diagnoses, there are often strong legal prohibitions that prevent public health authorities or other government officials from doing that same thing. So, now you've identified and defined your cases, you've gone out to the community or you've used administrative records to find those. Now, comes another tough part which is, what is the exposure of interest? In some ways this is similar to the case definition is, what is the thing that people are concerned about that might be linking these cases? Here it's often difficult to pinpoint a specific agent, and in many cases this is a conversation with the community, with community advocates, with experts, with others to think about what could be the source or sources that make this group of people vulnerable to this condition or set of conditions. Now, in some cases it's going to be something that's very well-defined. It might be for example that this is a group of individuals living in an area where there is no other obvious source of exposure, and there's something there like a leak into a water supply or something else which allows you to have a population that's otherwise unexposed, think about a specific exposure, link that to a specific outcome. That's in some ways the cleanest example, but unfortunately, that's not usually the case. Oftentimes, you have a situation where you've got people living in a community or in an area, and you have to think about what are all the things that could potentially link these people and to which they are exposed either in groups or individually, which could account for some of these health outcomes. So, for example, if you have a situation where you've got people who are living in a community and are concerned about some common health outcome, do they have a common exposure of some kind? Particularly, as I said if you can identify a single agent that one that's present that was not present before. Do they have something else in common? For example, are there more of them physically in one area of the community than in another area of the community, or do they have in common a workplace or something else? Is there another source that's potentially of concern like a hazardous waste site, or do you have a cluster or a concern about a cluster, and there's no obvious single source that's there? In situations like this, it can be very difficult to figure out what the most appropriate exposure is, and sometimes you want to wait a little bit and reserve judgment until you've looked just at that case counts, how many there are. As we'll discuss in another lecture on analysis, you may want to wait and look at the data for just your case counts and rates and then think about how, and whether to consider the question of exposure analysis.
