This presentation will address the rather broad categories of “Health Improvement” and “Public Health”
Objectives: In our objectives section we will identify the primary thrust of the presentation. Background: In the background we will address the two distinct topics pertinent to this presentation. The first being to establish what GIS is and its unique capabilities. The second being to speak towards research demonstrating the association between socioeconomic status and CVD disparity. Cardiovascular Disease and Socioeconomic Status Example: Next, we will walk through an example that demonstrates the utility of GIS in program development by looking at cardiovascular disease morbidity and socioeconomic status in Georgia. Conclusion: In our final section we will summarize the functionality of GIS in the example and the functionality of GIS in a broader sense to public health practitioners
My goal for you is not to teach you the technical side of GIS so that you can leave here knowing how to operate arcGIS. Rather, it is to demonstrate for how GIS might be incorporated in your own work and to get you thinking about the possibilities that are inherent in the process of easily and simply visualizing a disease, an area, risk factors, co-morbidities, et cetera ad infinitum and how this can be engaged in the decision-making process.
The database system in GIS consists of a database, where the data is stored in a logical manner, a database management system in the form of a computer software that provides the interface between an individual user and the database, and a relational database model. Spatially indexed data refers to data that is related to objects in space, such as objects (like a building, a car crash, a crime), lines (like roads, gas pipelines), or polygons (like states, counties, or census tracts). Procedures that might be operated upon a GIS include counting the number of cancer cases within a given county, constructing a 10 mile buffer around all trauma care hospitals to determine access, clipping only crashes within the state of Georgia from a national database of all crashes to create a new database, determine the rate of asthma cases by census tract.
We will be focusing on the first of these bullets. We will leave the functionality of GIS as a means of cluster detection, as demonstrated in the map to the right, and as a means of making regression analyses more specific as they incorporate spillover effects into the model itself for a later conversation.
This quotation was attained in response to a questionnaire in the United Kingdom regarding the uptake of GIS in the National Health Service. We, meaning those who utilize GIS on a daily basis, can’t just assume others will want to use GIS. We should be humble in the usage and promotion of GIS and in identifying what it is, especially with regards to program development and implementation. Yet at the same time we must acknowledge that the role of GIS in public health is constantly changing and that there is a need for more research to establish the context the GIS is most applicable.
GIS can be used to make program adjustments based not only on risk of the disease itself but also risk factors of the disease and other exposures such as socioeconomic position like we will be focusing on in our example. A program may incorporate these in their prioritization of potential targets or
Our interest in looking at the CVD-Socioeconomic status relationship is based on the growing body of evidence that links the two together. The line on top is the quintile with the lowest economic status, while the line on bottom at “1” is the quintile with the highest economic status. As the graph to the right demonstrates, areas of lower socioeconomic status are at greater risk of mortality for men – the same is true for women. This graph demonstrates how the disparity between the highest and lowest economic groups has actually widened a good deal over time, especially for those of the lowest economic group – the top line. In 1969, those in the most deprived quintile were only 30% more likely to die due to CVD, whereas in 1997, they were at nearly 80% higher risk of mortality due to CVD – for women, this figure is closer to 95% Moving forward, GIS could be used to address this ever-widening disparity in disease burden to mitigate the divide between those areas of the highest economic standing and those of the lower economic standing
Not only is it the disease itself which is associated with areal socioeconomic disparity, but so too are risk factors for the disease. The diagram to the right demonstrates the findings in that in a given city, with decreased economic deprivation, here illustrated with increasing house values, there is a concurrent decrease in current smoking amongst women – the same is true for men. Diez-Roux’s 1997 article, “Neighborhood Environments and Coronary Heart Disease: A multilevel analysis”
So, lets say that our objective is to increase the control of high blood pressure and cholesterol. We want to demonstrate how using GIS makes deciding which counties are in greatest need of controlling their high blood pressure and cholesterol easier and how GIS may impact the program by overlaying socioeconomic status data on the CVD data. Socioeconomic status is just one of a number of other factors that may be taken into account when designing a CVD intervention for a county – one may also be interested in co-morbid conditions like obesity and diabetes and several risk factors like smoking, nutrition, physical inactivity, or health access as means of decision-making. As the number of criterion increases so too does the decisionmakingGIS can help ease the process of not only understanding these dynamics both within an individual county and throughout the state but also communicating these dynamics.
(Read slide first) Area-based socioeconomic measures permit the routine monitoring of inequalities in health The Local Economic Resource Index and Poverty Prevalence have been shown to be sensitive to socioeconomic gradients in health White Collar Occupations: Management Occupations, except farm managers Business and financial operations occupations Professional and related occupations Sales and office occupations
(Go through slide first) Deduplication indicates that we counted an individual only once even if re-admitted in the same calendar year Though we had access to all diagnoses we utilize the principle diagnosis as that is intended to communicate the primary reason an individual visited the hospital in the first place.
(Read slide first) With these sort of background numbers it is easy to understand why there are programs already attempting to mitigate this issue and why there are others being developed to approach CVD reduction throughout the state.
(Read slide first) This single variable measure has been shown to perform as well as more complex measures of economic deprivation such as the Townsend Index while still maintaining the same qualities listed above
Point out which county is Dade (lowest/top left) and which is Heard (highest/7 th one down). This begins the prioritization of counties. Heard is a likely candidate for the CVD program intended to promote hypertension and cholesterol screenings as are the other counties in dark red.
Notice that this is based on “unemployment” so being a darker blue color is not a good thing.
After having taken into account the preceding indicators of socioeconomic status, the Local Economic Resource Index provides a meaningful way of understanding which counties have the most economic resources to pull from at a given time. As expected, Economic Resources are centralized around urban areas like Atlanta and Savannah while Rural Georgia counties are consistently among the most deprived counties.
In this map, being a darker shade of blue is not a good thing. One may notice that this map is essentially the reciprocal of the previous map depicting Local Economic Resources.
When comparing the difference in the hospital discharge rate for quintile 0, the least economically advantaged quintile, with quintile 4, the most economically advantaged quintile, we found that unemployment, education, and economic resources all had significant differences in the means of the two groups.
CVD morbidity and socioeconomic position are inversely related. CVD morbidity amongst counties in the lowest Local Economic Resource quintile is 20% higher as compared to counties in the highest Local Economic Resource quintile – the same is true for unemployment and education One has a couple of options with these results: First, one could extrapolate that a county with low socioeconomic status should have high CVD morbidity. How does this insensitive extrapolation get a program any closer to identifying target counties? Second, one could go back to the source dataset and identify the counties with the highest CVD morbidity and then find counties of the lowest socioeconomic status. But what about if a program wants to take other criteria into account, such as access to healthcare facilities, health insurance coverage, a co-morbid factor such as obesity? Then going back to a table seems tedious and out of touch with technological options that GIS can offer. This would traditionally be the endpoint of an Epidemiological analysis – charts, graphs, relative risks that establish a risk factor-disease association, but no maps and no spatially referenced findings being communicated.
Without the incorporation of spatially referenced data into the results what can an audience member say about a given county’s CVD burden or its socioeconomic status? What would happen if we went past these findings to help program planners and other stakeholders identify SPECIFIC areas in need of a program or those capable of having a program? By using spatially referenced data one is communicated to in a very straightforward and visual manner the value of the variables of interest and how they inter-relate. At this point we have seen the distribution of CVD morbidity and the distribution of socioeconomic position in each county. It is somewhat satisfying to have seen this, but we then need to ask ourselves how to integrate the two thoughts together in an effective way that expedites the decision-making process by not only demonstrating high priority counties but also unique traits of these counties pertinent to program design – in this case socioeconomic position.
The goal of utilizing 3-Dimensional graphics is to demonstrate where there is a confluence of both disease burden, risk factors, or other criterion like obesity, high cholesterol, current smoking, asthma, et cetera – any criteria that are important and able to be assessed from a programmatic decision-making perspective. This is where I view one of the strengths of GIS – presenting vast amounts of intelligible data in a concise format in place of presenting tables or portions of tables which are not effective communication devices in large scale or complex formats. (Begin arcScene after reading slide) (Focus on Poverty-CVD Morbidity 3-D map): In the West Central, or Columbus, Health District, we see one county in the least economically advantaged quartile and appears to have one of the highest CVD morbidity rates. This is Randolph County and it had a CVD morbidity rate of 1,882 age-adjusted deduplicated hospital discharges in 2008. It had a poverty rate of 24.3% and a Local Economic Resource Index score of ZERO, meaning that it was in the lowest quintile for unemployment (6%), education (11% have at least an Associate’s Degree), income (with a family median income of $30,000) and white collar employment (where 19% of the population was employed in a white collar occupation). The county was named after John Randolph, a Republican from Virginia who was also a descendant of Pocahontas. It has a population of almost 8,000 and covers 429.3 square miles and is home to one of the first pecan trees brought into GA. Up the GA line a bit you will reach another peak portraying an unusually high CVD morbidity rate. This is Troup County. However, this time it is one of the more advantaged counties. Troup has a CVD morbidity of 1,740 age-adjusted deduplicated hospital discharges in 2008. Its poverty rate was 16.4%. It has an LER score of 10. Breaking this down, it was in the 4 th highest quintile in white collar occupation at 31%. It was in the 2 nd quintile of unemployment at 5.8% (ahh, for the good ole days when 5.8% was considered among the worst counties). Troup was in the 4 th highest quintile for median income at $42,000 and also in the 4 th highest percent of population that has attained at least an Associates degree at 22%. Troup county is currently home to 68,000 residents. It was created out of a settlement with Creek Indians in 1827. Troup county is currently home to several industries, including textiles, product packaging, and batteries and is highly anticipating the construction of a KIA’s first American automobile manufacturing facility. If both of these counties could be chosen to receive the hypertension and cholesterol screening program, Randolph, the least economically advantaged county in GA may require more resources to achieve the goal of increasing the number of screenings. For example, as compared to Troup county, Randolph may benefit more from the addition of a comprehensive wellness center where individuals may go not just for free or reduced price hypertension and cholesterol screenings but also for exercise classes, healthy eating and cooking classes, smoking cessation counseling, and other health education opportunities. Whereas in Troup County, individuals may benefit more from simple media campaigns or other informational sessions, relying on their actual doctors to perform the screenings. Troup county residents may already have access to the healthcare and health facilities that residents of Randolph lack due to their greater economic resources to pull from. They may even have greater access to places like gyms and parks where they can safely exercise thus decreasing their physical inactivity rate. GIS allows us to easily find the target counties via the elevation. From here we can overlay any number of risk factors or other criterion that are important in program implementation. This allows us to communicate the data in a format intelligible to most audiences to guide the decision-making process. Ours is only one simple example for portraying the traits of a county’s disease burden and a single risk factor. One can repeat this process with any and all criterion by which one would want to make decisions on counties to target and which type of program to implement.
Given completion of the selection process we may have ended up with Houston and Irwin counties being ones receiving the intervention, one an economically advantaged county, the other a more disadvantaged county. Irrespective of whether programmatic personnel have chosen the more economically advantaged or disadvantaged county, a given program may then want to decide to prioritize target areas for the hypertension and cholesterol screening intervention based on CVD morbidity at a smaller subsection of the county – in our case, we chose to use a census block group. Knowing where burden is the highest in a chosen county may be useful in deciding where to further target resources. GIS allows us to swiftly and effectively move from the county level to this rather fine scale of block group. First, Houston county is one of the most economically advantaged counties in Georgia. However, as previously mentioned, it also has one of the highest CVD morbidities in Georgia. If the program doesn’t have much in the way of resources or is looking to somehow implement a program in one of the more advantaged areas, they may then narrow down the populations that they need to target to the block group level. A block group is smaller than a census tract and usually contains between 600 and 3000 persons – optimally about 1500. They are the smallest area for which sample information is available from the Census Bureau and are intended to reflect generally homogenous populations. Second, Irwin County is one of the most deprived counties in Georgia as it has an Local Economic Resource Index score of zero. Understanding that this is already a disadvantaged area without much in the way of economic resources, a program may want to focus on the block groups with the highest CVD morbidity to more directly address the disease. Houston County’s highest CVD morbidity rates occur seemingly randomly throughout the county whereas Irwin County’s seem concentrated in the Northeast side. Understanding these two aspects further enable program planners to more effectively target those populations suffering from the highest burden of CVD. In Irwin County, given the concentrated nature of the CVD morbidity, a program planner may come to the conclusion that if a comprehensive wellness center is to be built it may be most fruitful to place it so that these populations have greater access to it than those that may not have such a high burden. Houston county on the other hand, may benefit more from a more broad spectrum media campaign to increase awareness of hypertension and cholesterol screening and its benefits.
(Read through slide first) In the example we have used, GIS provided a means of understanding the relationship between a disease, CVD, and a risk factor thereof, socioeconomic status. It allowed us to quickly and simply identify where there was a confluence of high CVD morbidity and low socioeconomic status – presumably this would be the county or counties in need of the most help. Though the same conclusions could have been drawn using traditional database systems without the use of GIS, the data integration and visualization capabilities combined with the ability to spatially reference data allows for a more intuitive, less cumbersome means of comprehending and communicating which counties should be the targets of programs and which programs and health messages would be best suited to a particular county.
It is GIS’s ability to effectively portray the confluence of a given disease and its risk factors or co-morbidities that makes it useful to public health practitioners who are looking for a given area to target. The example provided is from the state government’s perspective on the distribution of funds for public health programs in a given county. This same approach could be useful at most any scale as in the maps portraying the block groups for Houston and Irwin counties. Sure, the same data can be understood in table format without GIS as the individual from UK likely does very well. But how does one then communicate this data to their audience? As my hypothetical program design consortium, would you have preferred me to present 159 observations for each of the 7 variables we have used in a table? Would that have been an effective means of internalizing and comprehending where there is a confluence of high CVD-low SES at the same time as noticing where there is also high CVD-high SES? This ability to take data from disparate sources and at multiple levels and then to communicate it in a way that allows others to quickly and easily determine the burden of disease on an area and other spatially referenced variables of importance is a tool that public health practitioners may find incredibly useful as they utilize data-based decision-making.
Life expectancy for those
What is the purpose of those two adjectives? Final comments Do you think you have answered sufficiently the question you raised initially: “ We haven’t got GIS. It isn’t a problem for us. Why is it a problem for you?” I think we can get the information you presented without drawing any map at all. See slides 34 & 35. The question, rather, is what does GIS add to that information? If you used CVD as an example, then you have to give clear examples of how GIS helps to make the right decision in priority setting and choice of intervention. What activities are included in the CVD prevention or health promotion programs? And which ones should be directed to which population group? How should a component of a program be delivered for one community differently? What are the common parameters used for program development or priority setting or resource allocation, etc and which ones are influenced by GIS or what additional dimension does GIS bring in?
One way of conceptualizing how to apply what we know of both CVD morbidity and Economic Resources is with the following. Have a very simplistic conceptualization of the CVD Morbidity-Local Economic Resources relationship: Green: Low Economic Resources, low CVD Morbidity Blue: High Economic Resources, low CVD Morbidity Yellow: High Economic Resources, high CVD Morbidity Red: Low Economic Resources, High CVD morbidity (Click to next slide)
One way of conceptualizing how to apply what we know of both CVD morbidity and Economic Resources is with the following. Have a very simplistic breakdown of the CVD Morbidity-Economic Resources relationship: Green: Low Economic Resources, low CVD Morbidity Blue: High Economic Resources, low CVD Morbidity Yellow: High Economic Resources, high CVD Morbidity Red: Low Economic Resources, High CVD morbidity We had previously started our research under the premise that CVD morbidity would be inversely related to areas of fewer economic resources. We had previously started our research under the premise that CVD morbidity would be inversely related to areas of fewer economic resources. Our findings also indicate that one would expect the most deprived counties to have 20% higher CVD morbidity as compared to the least deprived counties. Though this may be true in general, it is not an absolute. So, there are likely to be exceptions to this trend.
Green: low economic resource, low CVD – These are some of the lowest CVD morbidities in the dataset from Clay and Quitman counties, both also having some of the lowest economic resources Blue: high economic resource, low CVD – this is what one would extrapolate from research findings where Catoosa and Dade, of higher local economic resources, having low CVD morbidity (click to next slide)
Yellow: high economic resource, high CVD – in this conceptual box one may want to take notice as Jones and Houston counties are of the highest Local Economic Resources, but they also have some of the highest CVD morbidities Red: low economic resources, high CVD – again, this is what one would expect from the research as Marion and Twiggs have very little local economic resources and they have two of the higher CVD morbidities. Jones, Houston, Marion, and Twiggs are all worthy of interventions for hypertension and cholesterol screening as they have a CVD morbidity nearing 2% of the population. Given that one had to pick only one county, I am guessing most of us would pick the county with the least amount of economic resources of these four – Marion. But, what if one could choose several counties for the intervention? One would then have to answer questions regarding program design as the program itself, though having the same objective, is likely to differ in a county of exceptionally high Local Economic Resources as compared to a county of exceptionally low Local Economic Resources
Geographic Information Systems for Resource Allocation
Geographic Information Systems for Resource Allocation Presentation to Georgia Public Health Association April 12, 2011 Michael Bryan, Chronic Disease Epidemiologist
DCH Mission ACCESS Access to affordable, quality health care in our communities RESPONSIBLE Responsible health planning and use of health care resources HEALTHY Healthy behaviors and improved health outcomes
DCH Initiatives FY 2011 FY 2011 Continuity of Operations Preparedness Customer Service Emergency Preparedness Financial & Program Integrity Health Care Consumerism Health Improvement Health Care Transformation Public Health Workforce Development
Outline <ul><li>Objectives </li></ul><ul><li>Background </li></ul><ul><li>Cardiovascular Disease and Socioeconomic Status Example </li></ul><ul><li>Conclusion </li></ul>
Objectives <ul><li>To demonstrate the use of GIS for program development and health message targeting </li></ul><ul><ul><li>To visualize and describe the spatial relationship between county-level socioeconomic indicators and cardiovascular disease morbidity in Georgia </li></ul></ul><ul><ul><li>To visualize the distribution of CVD morbidity on the block group level </li></ul></ul>
Number of Farms by County, Georgia 1997 Without GIS Visualization Capabilities County No. Farms County No. Farms County No. Farms County No. Farms County No. Farms County No. Farms Rabun 122 Gwinnett 303 Hancock 103 Muscogee 39 Appling 494 Decatur 335 Towns 121 Barrow 361 Butts 148 Effingham 203 Randolph 119 Grady 462 Fannin 151 Polk 344 Heard 160 Bleckley 221 Chatham 42 Thomas 421 Murray 238 Paulding 218 Spalding 193 Marion 147 Turner 230 Seminole 183 Whitfield 325 Cobb 128 Glascock 76 Candler 264 Ben Hill 159 Charlton 75 Catoosa 215 Oglethorpe 319 Jefferson 356 Chattahoochee 13 Worth 406 Lowndes 373 Union 256 Clarke 80 Burke 346 Macon 282 Wayne 276 Echols 67 Walker 478 Wilkes 298 Washington 327 Treutlen 157 Coffee 656 Camden 46 Dade 175 Lincoln 163 Meriwether 257 Dodge 491 Clay 56 Brooks 430 Gilmer 267 DeKalb 46 Troup 221 Schley 91 Irwin 288 Habersham 407 Oconee 305 Pike 252 Pulaski 161 Bacon 324 White 284 Walton 493 Lamar 188 Taylor 196 Lee 157 Lumpkin 198 Haralson 260 Monroe 179 Toombs 401 Dougherty 139 Stephens 188 Morgan 390 Baldwin 137 Montgomery 252 Calhoun 122 Gordon 535 Carroll 702 Jones 157 Tattnall 589 Tift 359 Dawson 160 Douglas 107 Screven 325 Wheeler 176 Pierce 379 Chattooga 278 Rockdale 102 Wilkinson 88 Dooly 259 Early 279 Floyd 437 Greene 198 Upson 185 Evans 183 Berrien 399 Pickens 194 Newton 260 Jenkins 248 Bryan 61 Ware 274 Franklin 699 Taliaferro 55 Bibb 149 Webster 76 Baker 131 Hall 666 Columbia 169 Twiggs 98 Stewart 77 Mitchell 464 Hart 460 McDuffie 217 Talbot 111 Sumter 314 Atkinson 196 Banks 446 Clayton 54 Harris 207 Telfair 271 Brantley 207 Bartow 400 Henry 327 Crawford 123 Wilcox 273 McIntosh 24 Cherokee 493 Warren 134 Emanuel 441 Liberty 43 Cook 226 Forsyth 434 Fayette 184 Johnson 288 Crisp 213 Colquitt 634 Jackson 719 Richmond 106 Laurens 688 Long 64 Miller 251 Elbert 320 Jasper 185 Peach 157 Quitman 17 Clinch 93 Madison 622 Coweta 316 Houston 249 Jeff Davis 220 Lanier 92 Fulton 257 Putnam 152 Bulloch 524 Terrell 174 Glynn 36
Number of Farms by County, Georgia 1997 With GIS Visualization Capabilities
Research Question <ul><li>Do Geographic Information Systems help guide program development and health message targeting? </li></ul>
What is GIS? <ul><li>A “database system in which most of the data are spatially indexed and upon which a set of procedures are operated in order to answer questions about spatial entities in the database.” (Antenucci 1991) </li></ul>
GIS Defined <ul><li>Database System </li></ul><ul><ul><li>Database </li></ul></ul><ul><ul><li>Database management system (DBMS) </li></ul></ul><ul><ul><li>Relational Database Model </li></ul></ul><ul><li>Spatially Indexed </li></ul><ul><ul><li>Data related to items in space, like objects, lines, or polygons </li></ul></ul><ul><li>Procedures </li></ul><ul><ul><li>Ways to manipulate spatially indexed data in database system </li></ul></ul>
Questions for GIS <ul><li>Where along I-85 is the highest fatal crash rate using 500 meter segments of the interstate as observational unit? </li></ul><ul><li>What was the distribution of the Chlorine gas plume that occurred in Conyers, GA in 2004 beginning at 5am and ending at 5pm in 10 minute increments? </li></ul><ul><li>Is the incidence of Lyme Disease in South Georgia associated with urbanization? </li></ul><ul><li>Is county obesity prevalence associated with green space acreage? Sidewalk length? </li></ul>
What does GIS do? <ul><li>Capture Data </li></ul><ul><ul><li>Identify objects and enter data on these objects </li></ul></ul>Cardiovascular Disease (CVD) Discharges, Georgia 2008
What does GIS do? <ul><li>Integrate Data </li></ul><ul><ul><li>Combine data from different sources and/or different scales </li></ul></ul>County Population Data from US Census Bureau County CVD Hospital Discharges from GA Hospital Association County Database
What does GIS do? <ul><li>Manipulate Data </li></ul><ul><ul><li>Process data in database </li></ul></ul>County Population CVD Cases County Population CVD Discharges CVD Morbidity County Database
What does GIS do? <ul><li>Produce Maps </li></ul><ul><li>Produce Graphs and Tables </li></ul><ul><li>Produce Reports </li></ul><ul><li>Geographically-based analysis </li></ul>
Applications of GIS <ul><li>Targeting resources towards particular groups </li></ul><ul><li>Planning locations of health facilities and programs </li></ul><ul><li>Determining catchment areas and target populations </li></ul><ul><li>Creating health profiles </li></ul><ul><li>Epidemiological research and analysis </li></ul><ul><li>Assessing health needs to provide health services </li></ul>
“ We haven’t got GIS. It isn’t a problem for us. Why is it a problem for you?” <ul><li>Removes technology and tools available </li></ul><ul><ul><li>To investigate impacts of exposures to human health </li></ul></ul><ul><ul><li>To monitor diseases and their risk factors </li></ul></ul><ul><ul><li>To determine health inequalities </li></ul></ul><ul><ul><li>To communicate with others </li></ul></ul>
Geographic Information Systems and Health Interventions <ul><li>GIS helps identify areas or populations at risk of disease </li></ul><ul><li>GIS helps relate disease risk to potential areal risk factors </li></ul><ul><ul><li>Socioeconomic position </li></ul></ul><ul><ul><li>Amount of tobacco advertising </li></ul></ul><ul><ul><li>Availability and cost of healthy food </li></ul></ul><ul><ul><li>Availability and quality of public spaces </li></ul></ul><ul><ul><li>Sense of safety or crime </li></ul></ul><ul><ul><li>Exposure to chronic stress </li></ul></ul><ul><ul><li>Sources of social support </li></ul></ul>
Geographic Information Systems and Health Interventions <ul><li>Prioritize Target Populations </li></ul><ul><li>Adjust Intervention </li></ul>Disease Risk Risk Factors Other Exposures Target Population Intervention Type
Geographic Information Systems and Health Interventions Number of Criterion Complexity Utility of GIS
Socioeconomic Disparity and CVD Burden <ul><li>Burden of CVD greater in areas of lower socioeconomic position </li></ul><ul><li>Socioeconomic inequality in burden of CVD is increasing with time </li></ul>Source: Singh (2002)
Socioeconomic Disparity and CVD Risk Factor Burden <ul><li>Persons living in more deprived areas have </li></ul><ul><ul><li>Increased risk of obesity </li></ul></ul><ul><ul><li>Increased smoking </li></ul></ul><ul><ul><li>Increased physical inactivity </li></ul></ul>Diez Roux (1997)
Socioeconomic Disparity and Health Interventions <ul><li>More deprived areas may be less susceptible to prevention efforts </li></ul><ul><ul><li>Lower health knowledge </li></ul></ul><ul><ul><li>Lower probability of healthy behavior change </li></ul></ul><ul><ul><li>Less exposure to prevention messages </li></ul></ul><ul><ul><li>(Benjamin-Garner 2002; Bartley 2000) </li></ul></ul>
Cardiovascular Disease Morbidity and Socioeconomic Indicators Example
CVD Program <ul><li>Objective: Increase hypertension and cholesterol screening rate </li></ul><ul><ul><li>Target populations of highest CVD burden </li></ul></ul><ul><ul><li>Utilize socioeconomic status in program design </li></ul></ul>
Data Sources <ul><li>GA Hospital Association </li></ul><ul><ul><li>CVD Morbidity </li></ul></ul><ul><li>US Census Bureau </li></ul><ul><ul><li>Education </li></ul></ul><ul><ul><li>Occupation </li></ul></ul><ul><ul><li>Income </li></ul></ul><ul><li>Bureau of Labor Statistics </li></ul><ul><ul><li>Unemployment </li></ul></ul><ul><li>US Department of Agriculture </li></ul><ul><ul><li>Poverty </li></ul></ul>
Variables <ul><li>Age-Adjusted Cardiovascular Disease Morbidity </li></ul><ul><li>Local Economic Resource Index </li></ul><ul><ul><li>Unemployment Rate </li></ul></ul><ul><ul><li>Percent with at least Associate’s Degree </li></ul></ul><ul><ul><li>Family Median Income </li></ul></ul><ul><ul><li>Percent of working population in white collar occupation </li></ul></ul><ul><li>Poverty Prevalence </li></ul>
Cardiovascular Disease Morbidity <ul><li>Age-adjusted to 2000 US Standard Population </li></ul><ul><li>Deduplicated 2008 Hospital Discharges </li></ul><ul><li>Principle Diagnosis </li></ul><ul><ul><li>ICD-9 codes 390-434 and 436-448 </li></ul></ul><ul><ul><li>Ischemic Heart Disease </li></ul></ul><ul><ul><li>Hypertensive Heart Disease and Hypertension </li></ul></ul><ul><ul><li>Stroke </li></ul></ul><ul><ul><li>Rheumatic Fever and Chronic Rheumatic Disease </li></ul></ul>
Burden of Cardiovascular Disease Morbidity, Georgia 2008 <ul><li>145,000 total hospitalizations due to any CVD </li></ul><ul><li>Average length of stay was 5 days </li></ul><ul><li>Average charge per hospitalization was $35,800 </li></ul><ul><li>Total hospital charges were $4.9 billion </li></ul><ul><li>Direct healthcare cost and indirect cost estimated at $11.7 billion </li></ul>
Local Economic Resource Index <ul><li>Summary Index of 4 measures: </li></ul><ul><ul><li>Percent of Working Population in White Collar Occupation </li></ul></ul><ul><ul><li>Unemployment Rate </li></ul></ul><ul><ul><li>Median Family Income </li></ul></ul><ul><ul><li>Percent with Associate’s Degree or greater </li></ul></ul><ul><li>Variables categorized into quintiles </li></ul><ul><li>Quintiles assigned scores from 0 to 4 </li></ul><ul><li>Scores of 0 represent the most economically disadvantaged group while scores of 16 represent the most economically advantaged group </li></ul>
Poverty <ul><li>Percent of persons living below the US poverty line </li></ul><ul><ul><li>Based on income thresholds that vary by family size and composition </li></ul></ul><ul><li>Captures economic deprivation </li></ul><ul><li>Meaningful across regions and time </li></ul><ul><li>Easily understood and interpretable </li></ul>
Percent of Population in Poverty, Georgia 2007 <ul><li>Range: 5.2% - 36.2% </li></ul><ul><li>Median: 18.3% </li></ul><ul><li>Mean: 18.6% ( σ =6.4%) </li></ul>
Socioeconomic Disparity and CVD Morbidity*, Georgia *Age-Adjusted Hospital Discharges per 100,000 population Significant
Summary of Visualization of CVD and Socioeconomic Indicators <ul><li>Those in the most economically disadvantaged Local Economic Resource quintile have a 20% higher morbidity than those in the most economically advantaged quintile </li></ul>Socioeconomic Status CVD Morbidity
What does GIS give to programs? <ul><li>Communication tool for themselves and their stakeholders </li></ul><ul><li>Method of incorporating several variables of import </li></ul><ul><ul><li>Facilitates data-based decision-making </li></ul></ul>
3-D graphics to visualize CVD Morbidity and Socioeconomic Indicators Simultaneously
Cardiovascular Disease Morbidity by Block Group, Houston and Irwin County 2008 Houston County Irwin County
GIS as Tool for CVD Program <ul><li>Visualize the burden of CVD morbidity and the economic position for each county </li></ul><ul><li>Visualize the distribution of both disease burden and economic resources for all counties throughout Georgia </li></ul>
GIS as Programmatic Tool <ul><li>GIS provides: </li></ul><ul><ul><li>Simple way to understand the relationship between disease burden and disease risk factors </li></ul></ul><ul><ul><li>Means of incorporating any spatially referenced variables of interest </li></ul></ul><ul><li>Technological advances allow for a fine scale picture of the health climate </li></ul><ul><ul><li>Can target health programs accordingly </li></ul></ul>
References <ul><li>Armstrong D et al. “Community occupational structure, medical and economic resources and coronary mortality among US blacks and whites, 1980-1988.” Annals of Epidemiology 1988; 8:184-191. </li></ul><ul><li>Bartley M et al. “Social distribution of cardiovascular disease risk factors: change among men in England 1984-1993. J Epidemiol Community Health 2000; 54; 806-814. </li></ul><ul><li>Benjamin-Gamer R et al. “Sociodemographic differences in exposure to health information. Ethn Dis 2002; 12; 124-34. </li></ul><ul><li>Curtis AJ and Lee WA. “Spatial Patterns of diabetes related health problems for vulnerable populations in Los Angeles.” International Journal of Health Geographics 2010; 9(43); 1-10. </li></ul><ul><li>Diez-Roux AV et al. “Neighborhood Environments and Coronary Heart Disease: A multilevel analysis.” American Journal of Epidemiology 1997; 146(1); 48-63. </li></ul><ul><li>Diez-Roux AV et al. “Neighborhood of residence and incidence of Coronary Heart Disease.” New England Journal of Medicine 2001; 345 (2); 99-106. </li></ul><ul><li>Ezzati M et al. “The Reversal of Fortunes: Trends in County Mortality and Cross-County Mortality Disparities in the United States.” PLoS Medicine 2008; 5(4); 0557-0568. </li></ul><ul><li>Gesler WM et al. “Using mapping technology in health intervention research.” Nursing Outlook 2004; 52; 142-146. </li></ul><ul><li>Lawlor DA et al. “Life-Course Socioeconomic Position, Area Deprivation and Coronary Heart Disease: Findings From the British Women’s Heart and Health Study.” American Journal of Public Health 2005; 95(1); 91-97. </li></ul><ul><li>Lyratzopoulos G et al. “Deprivation and trends in blood pressure, cholesterol, body mass index and smoking among participants of a UK primary care-based cardiovascular risk factor screening programme: both narrowing and widening in cardiovascular risk factor inequalities.” Heart 2006; 92; 1198-1206. </li></ul><ul><li>Singh GK. “Area Deprivation and Widening Inequalities in US Mortality, 1969-1998.” American Journal of Public Health 2003; 93(7); 1137-1143. </li></ul><ul><li>Singh GK and Siahpush M. “Increasing inequalities in all-cause and cardiovascular mortality among US adults aged 25-64 years by area socioeconomic status, 1969-1998.” International Journal of Epidemiology; 31; 600-613. </li></ul><ul><li>Smith DP et al. “Re(surveying the uses of Geographical Information Systems in Health Authorities 1991-2001.” Area 2003; 35(1); 74-83. </li></ul><ul><li>Sundquist J et al. “Cardiovascular risk factors and the neighborhood environment: a multilevel analysis.” International Journal of Epidemiology 1999. 28; 841-845. </li></ul>
Acknowledgements <ul><li>Rana Bayakly, MPH, Chronic Disease, Healthy Behavior, Injury, Environmental Epidemiology Director </li></ul><ul><li>Lydia Clarkson, MPH, Cardiovascular Disease Unit Lead </li></ul><ul><li>Jim Steiner, Data Manager </li></ul><ul><li>All others who helped in any way </li></ul>
Thank You Michael Bryan, MPH [email_address] (404) 463-3748
Socioeconomic Disparity <ul><li>CVD burden decreasing more slowly in areas of lower socioeconomic position </li></ul><ul><ul><li>Life expectancy rising more slowly </li></ul></ul><ul><ul><li>Mortality/morbidity decreasing less rapidly in men </li></ul></ul>Top 2.5% Bottom 2.5%
White Collar Occupations <ul><li>Management Occupations, except farm managers </li></ul><ul><li>Business and financial operations occupations </li></ul><ul><li>Professional and related occupations </li></ul><ul><li>Sales and office occupations </li></ul>
Socioeconomic Indicators Indicator Range Median Mean( σ ) High Education 7.6% - 46.1% 15.6% 18.2% (8.1%) White Collar 29.9% - 72.3% 45.2% 47.2% (8.5%) Median Family Income $27,232 - $78,853 $38,463 $40,411 ($9,485) Unemployment Rate 3.0% - 9.5% 4.9% 5.1% (1.2%) Poverty 5.2% - 36.2% 18.3% 18.6% (6.4%)
GIS as Programmatic Tool <ul><li>Impact on programmatic decision-making </li></ul><ul><ul><li>Given limited resources, should the program target counties that have higher levels of economic resources or counties that have lower levels economic resources? </li></ul></ul><ul><ul><li>Should different programs and messages be implemented in areas based on a better comprehension of an area’s economic resources and disease burden? </li></ul></ul>
The Health Message Conundrum CVD Morbidity Economic Resources Low Low High High
The Health Message Conundrum CVD Morbidity Economic Resources Low Low High High
Examples of CVD Morbidity by Local Economic Resource (LER) Index *Age-Adjusted Deduplicated Hospital Discharges per 100,000 population County LER Quintile CVD Morbidity* Clay 0 417 Quitman 0 253 Catoosa 4 316 Dade 3 210 Jones 4 1800 Houston 4 1758 Marion 0 1799 Twiggs 1 1749
Examples of CVD Morbidity by Local Economic Resource (LER) Index *Age-Adjusted Deduplicated Hospital Discharges per 100,000 population County LER Quintile CVD Morbidity* Clay 0 417 Quitman 0 253 Catoosa 4 316 Dade 3 210 Jones 4 1800 Houston 4 1758 Marion 0 1799 Twiggs 1 1749