Rural Health Issues and Their Investigation in a GIS Environment

The previous chapters have provided a background in factors related to infant and maternal health and the various ways these individual and neighborhood characteristics can be studied in a GIS environment. Though rich in information for approaching investigations in urban areas, this book has yet to delve into these issues from a rural perspective, wherein some of the salient variables differ and certainly these data require additional understanding for use with GIS. Therefore, this chapter introduces some of the topics that could be included in health investigations in rural areas, along with some of the caveats for working with and interpreting the GIS results.

GIS and Spatial Analysis

n the opening chapters, GIS was broken into four general components, one of which was the spatial analysis of data. This is probably the least utilized of all GIS functions outside of an academic environment. A point that is often missed when discussing GIS is that the technology often exceeds the capabilities of the user. This is especially true if the user has not received any academic training in spatial data and GIS use. In Chapter VI a more sophisticated overview will be presented of the latest spatial analysis techniques along with examples of their implementation. Although the number of “spatially” trained scientists continues to grow, there is still a gap between the number of available skilled GIS modelers and the community programs needing GIS analysis. This chapter is designed to provide a stopgap approach, using more simple spatial statistical approaches that can be applied to gain a reasonable first insight into a birth outcome surface.

The Geography of Health Risks

This book has so far provided an introduction to GIS in terms of its use as part of a community health program. Subsequent chapters will describe a selection of more detailed GIS techniques and approaches. This chapter can be considered as an interlude, an attempt to set the scene by painting a backcloth of risks facing many city neighborhoods and the pregnant women living in them. Of course, at the risk of repeating oneself, there have been whole books written about single risk factors, so the task of compressing all risks into a single chapter is extremely difficult. This is made even harder because many of the situations described are not directly related to pregnancy outcomes, but instead are classed as neighborhood risks, contributing to the overall vulnerability of a person living in these environments. Although some may question why not concentrate solely on pregnancy-related risks, a more holistic understanding of the social environment can help place context into data, a movement away from the earlier criticized “structural functionalism” approach (Litva & Eyles, 1995). At this point it might be worth briefly mentioning that debate continues within the field of medical geography as to the degree in which pure analytical approaches ignore the social relevance of actions (a political economy approach), or how an individual’s experience shapes his or her actions (Dorn & Laws, 1994). For a review of these critical literatures in association with infant mortality see Gesler, Bird, and Oljeski, (1997).

An Introduction to GIS (All Things Data)

One of the purposes of this book is to introduce community health groups to the potential of GIS, a technology that can help in understanding the spatial landscape of prenatal risk. Therefore, one of the first steps is to provide a brief overview of GIS. This introduction will be split over the next two sections, with this chapter focused on data issues associated with using a GIS, while the next presents an introduction to the functions of a GIS that make it so powerful: the ability to analyze and visualize spatial data. These two chapters are not meant for experienced GIS users (though even for these a few points and references from the non-geographic literature may prove to be useful). It is also not meant to be a comprehensive introduction to the science; there are several other excellent texts serving that need. These next two chapters are meant to give a basic understanding, and inform enough to encourage the adoption of a GIS approach. Most people reading this book will probably be using a vector GIS. There are, however, two basic GIS formats, raster being the other type. Raster GIS is best suited for surface or complete coverage data (for example, vegetation cover) because the spatial surface is transformed into a grid, with each cell or pixel containing a relevant geographic attribute (such as 1 = water, 2 = forest, etc.). These pixels are fixed in space, allowing multiple layers at a single location to be compared and analyzed. Vector GIS, which is a more useful GIS format for the type of investigation likely to be performed by a health unit, contains points, lines, and areas. Unlike in a raster GIS, each spatial object has its own geography.

Spatial/Temporal Stability in Neighborhoods of Risk

Four years have been spent analyzing infant health data for Baton Rouge. As with working with any dataset, one gets a feel for the data. One such feeling was that the population being studied was mobile. The first indication came when trying to categorize the different neighborhoods in which Healthy Start program participants lived. The main reason for performing such an analysis was to gain an insight into any program participant entering the Healthy Start program, both in terms of her own previous birth outcomes and those of the neighborhood in which she lived (this will be revisited in the chapter on the Baton Rouge Healthy Start). In order to do this, one requirement was to trace pregnancy histories. As the birth and death certificate data did not come with unique identifiers, an alternative common field between birth certificates was needed. Originally, it was thought that a combination of street address, mother’s date of birth, and the date of birth of the previous child would suffice. It soon became obvious that mothers living inside the five zip code region served by the Healthy Start project were mobile. A more in-depth investigation into the degree of this mobility was needed, as there could be implications both in terms of GIS analysis, and the subsequent targeting of outreach into those identified neighborhoods.

An Introduction to GIS (All Things Spatial)

The last chapter provided an introduction to two of the key components in a GIS: getting the data in, and then manipulating them to answer questions. This chapter considers how these data can be visualized and analyzed. The analysis section in this chapter will be fairly simplistic, as this topic is revisited again in Chapters V and VI with the introduction of increasingly more sophisticated measures. The chapter will end with a discussion of an important topic integral to urban area analysis but which is often missed when discussing the GIS: What exactly is a neighborhood and how should it be spatially defined?

Bioterrorism, Pregnancy, and Old White Men

By the title you might think this chapter is a departure from the general theme of this book. I beg your indulgence as I plead my case, because pregnant women as a cohort have been tragically forgotten in the hazards and disaster literature. Although mapping of vulnerable populations is a recognized approach, I argue (and frequently have) that we need to reconsider our disaster mitigation, response, and recovery plans, directing attention toward neighborhoods with high proportions of high-risk pregnant mothers. I will argue my case over the course of this chapter and at the same time suggest a potential funding strategy that can tap into Homeland Security funding. This chapter will focus on one particular disaster, a bioterrorist (BT) attack, and suggest how community health units could leverage funding to start their own GIS in the form of a syndromic surveillance system.

Advanced Spatial Analysis

The last chapter presented several ideas of how to perform relatively simple forms of spatial analysis. Many of these approaches, though insightful, have been superceded by more advanced analytical techniques. This chapter will present a few of these approaches, namely methods of clustering, interpolation, and spatial association. Other concepts will also be addressed, such as spatial autocorrelation and the measures that can be used to find spatial clusters of significantly high (hot spots) or low (cold spots) values. Two additional cluster methods will also be discussed, these being nearest neighbor hierarchical clustering and the spatial filter. Kernel density interpolation will be introduced as the interpolation method for discrete incident locations. A discussion about spatial regression analysis will conclude this chapter. The analyses and examples shown in this chapter will again be based upon linked birth and death certificate data for East Baton Rouge Parish.

Creating the Baton Rouge Healthy Start GIS

I recently gave a presentation to an Introduction to Gender and Minority Studies class at Louisiana State University. This was an interesting experience, as most of my talks tend to lean on the technical aspects (meaning GIS) of how to solve the problem of racial disparity in birth outcomes in the city. This audience was nontechnical and more interested in the injustices facing the poor (and more often than not African Americans) in the city. I would not have been comfortable giving this talk a couple of years previously, but 4 years of being attached to the Baton Rouge Healthy Start, writing continual updates and reports, applying for new grants, and attending all the associated meetings (e.g., evaluation meetings, project area committee meetings, Fetal and Infant Mortality Review (FIMR) meetings) have given me a reasonable insight into the causes of the problems as well as the distribution of dots on a map.

Explaining the Geography of Infant Health

During a seminar presentation to the Geography and Anthropology Faculty at Louisiana State University, a series of summary statistics were presented concerning the racial inequality in infant health outcomes in Baton Rouge. The numbers were quite startling, spurring one colleague to question the accuracy of the findings, as, in their words, “that would put us at developing world levels.” This sentiment was echoed by a National Science Foundation reviewer (I suspect from the United Kingdom) who read one of our proposals and expressed disbelief. The United States is a modern developed country, and yet in geographical pockets the infant mortality rate is alarmingly high. For some neighborhoods, between 1996 and 1998, in East Baton Rouge Parish the infant mortality rate was consistently above 40/1,000, reaching its highest rate in 1998 of 70/1,000. This means if 1,000 babies were born in this 0.25-mile neighborhood, 70 would die in the first year of life. Obviously, there is a problem. This book will investigate a host of ways to consider, and hopefully move towards solving, problems such as these using a Geographic Information System (GIS). In this chapter we will provide an introduction to the main themes of the book, namely negative pregnancy outcomes and the role geography and geographic techniques can play in their improvement.