I'm now going to move on to the presentation of surveillance data in the context of maps. Now, most people again have looked at surveillance data or data represented by maps. The format are graphic representations of data using location and geographic coordinates. They represent clear and quick approaches for grasping data for people that are familiar with the geography. The types of maps broadly include spot maps and choropleth maps. Now spot maps have a lot of different names but most commonly also referred to as point pattern data. This is really where you place a dot or another symbol on a map where the health condition occurred or where it exists. Public health practitioners can look at maps like this and generate hypotheses and focus investigations. But importantly, it does not include a population size, so this is not a rate. Here we're looking at a map of the United States and Puerto Rico, looking at HIV aids cases through to 2005 which at that point represented a total of one million cases. What we can see in this point pattern data are that cases of HIV are predominantly within large urban centers across the United States. So we can see Los Angeles, San Francisco, Denver, New York, parts of Florida, really highlighted as epicenters of the HIV epidemic across the United States. So it allows public health practitioners working at the federal level to really start focusing prevention and treatment programs in those settings. Choropleth maps are also commonly used and these are also known as shaded or area maps. The format is that you use a shade, or a hatching, or a color to represent range-graded values. We normally use these to depict rates by region, but importantly there's clear ecologic fallacy at place. Which is to say that even if something is happening at the level of the state, it may not apply to an individual within that state. It also gives the false impression of abrupt change in the rate at the boundary, when really that is just a reporting bias and an artifact of the systems that are used to present these data. In reality we would expect a much more consistent change across any geographies. Here we're looking at four maps of the United States including two focused on obesity and two focused on diabetes. In terms of obesity, we're looking at the prevalence of obesity in 1991 as well as in 2001 with the different colors ranging from where there were no data, all the way up to where greater than 25 percent of the populace in that state was affected by obesity. Clearly we can see that there has been a great increase in the amount of people that are obese across the United States with really a focus in certain parts of the United States, again allowing public health practitioners to develop hypotheses and consider programs in response. Similarly when we look at diabetes, we're looking at data from 1991 to 2001 and we also see great increases in diabetes across the United States. However, there are significant parts that are disproportionately affected, and again allows practitioners to really consider appropriate strategies for those settings. Choropleth maps can also be presented in the context of large-scale surveillance data. These are data from the Behavior Risk Factor Surveillance System, and one can go on the CDC website and actually use this system to get an assessment of a lot of different health indicators. Here we looked at the adults aged 18-64 who had any kind of health care coverage, again, pre-Obama care. So as we can see here, there are clearly parts of the country that have greater access to healthcare coverage than other parts of the country. A map like this just allows you at a very top-line level to have an understanding at the state level as to what is happening in terms of health care coverage. But again it's really important to keep in mind ecologic fallacy. That is not to say that individuals in Texas do not have health care coverage, it is to say that overall the total populace is less likely to have health care coverage in that state. Another tool that is important is called WISQARS, and this is an interactive system for queries of injury surveillance data. These data that I'm presenting today are older from 2000-2006, but one can use this system by state, by zip code, by census tract to really get an idea of a lot of different injury-related data. These injuries can be both intentional and unintentional injuries. Here I've given an example that include death rates per 100,000 population for firearms. Now firearms can include both intentional and unintentional injuries. In terms of intentional injuries, these can be homicides as well as suicides. Here I've only presented homicide data, all races, all ethnicities, both sexes. What you can see here importantly is that up until 2006, data were only available as it related to firearm based homicides in a certain number of census tracks across the United States, and data are either suppressed or undefined in large parts of the United States. Now if you're interested in using the WISQARS system, there'll be a link on the lecture page to the WISQARS system where you can really visualize a broad range of injury-related data. I think it's important again because it allows you to have an understanding of what data are already available if this is a particular area of inquiry for you. Now here I've summarized in a table what are the appropriate types of graphs, charts, and maps that are appropriate for different types of data. This is really more of a resource for yourself, but it's important to keep in mind as you're thinking about different types of data to really be deliberate about the type of visualization that you use for those data. I will also note that there are increasingly new tools for data visualization that have been developed including tableau, Power Business Intelligence, among others because of the importance of effectively visualizing large data. As we continue to move towards precision medicine, precision surveillance; the use of big data and visualization, and synthesis approaches to big data are becoming increasingly important.
