Mapping Disease in General – Part I

Map Production List

Viewing notes

Several forms of data were mapped. The most common are raw data, independent prevalence (IP, with n compared to total popualtion for area reviewed), and grid of the raw data.

All maps are based on spatial data attached either directly to the cases and then zip codes, or to an address that is either specific or general (street name only, town, PO Box) in format and linked to zip codes. The final product of this process is the grid map, in which a grid cell is used to normalize the data spatially, allowing for accurate isoline and #D projections of the results. These grid maps were designed to improve upon some detailed math analysis I was engaged in, by also normalizing both demographically and prevalence-risk based results.

Due to the nature of the end products (the print outs), the zip code centroid datamaps are easiest to read. Compared with the others, they are not overwhelmed by color layers and tend to be less straining on the eyes. But due to the irregularity of shape, size form and even incongruity (bisected zip polygons of zip code polygons), zip codes must always be considered as being potentially error prone with analyses requiring exact location or exact/true area definition (the right place-zip code association).

Common problems encountered in the past include lack of reliance on zip code areal/centroid data due to the inaccracy of the placement of a centroid. High irregular zip code tracts can have a centroid located outside the zip code polygon. During the early address matching years, using address matching/geocoding software produced by three different companies would often have three different lat-longs identified for certain point data. In one of my earliest such projects carried out from 2002 to 2005, a total of 3500 cases had to be mapped. The data included “complete” addresses provided by the state public health database provider. Address matching tools at the time produced only a 50 to 55% match. It took several months to use Google Maps and field investigations to manually geocode the remaining cases. [Some maps of this pre and post-manual geocoding may in fact be posted; see Oregon Toxic Release Sites and cancer cases project.]

Larger Area Depiction for the Map Below

Notice the 80238 Zip Code, a split zip code with its centroid over Interstate 70, US Rte 36, the Zip Code for which is 80239. 80239 is a rural farming/agriculture tract that is 5 times larger than the sub-urban residential tract 80239 of Stapleton.

In general, it is safe to say that the use of grids to redefine zip code data results in greater spatial accuracy. The only problem it leaves us with is population density information. The use of grid cells instead of zip codes tracts, when the cell is larger than the average zip code tract, results in the merging of zip code data, making the final end populations more congruent in terms of numbers and validity. This enhances the results and outcomes of the grid maps enabling them to produce more accurate isoline maps of results. In addition to this improvement in numerics is the placement of these data, which as stated earlier is more conducive to producing fairly accurate contour maps of the results. This means that the isoline maps produced are more accurate.

In spite of these positive outcomes for grid mapping data, there is the simple problem of readability. The grid maps appear busy at times, and the color coding is sensitive to changes that are made in base data. These minor problems still need to be perfected at this end. A few examples of this are provided in the maps series below.

Concerning the methodology employed for the examples presented, a crude, on-the-fly method is in use for grid mapping most of the data. This means that the grids produced make use of just the zip code lat-long data compiled into gridcells based on where the zip code centroid is placed. This point represents all of the data for that zip code tract enven though the tract is partially overlaying another cell. This is a crude way of interpreting zip code data, but in the long run produces little error due to the diffuseness and randomness of these poorly located centroids/events relative to theoverall dataset. (This is explained in detail on the grid map page for this section.)

This latter error producing method can be automatically corrected for using ArcGIS. Since ArcGIS is not the tool employed for this mapping, no such correction process is in use. Therefore, this remains the major source for error when interpreting these types of maps.

Since the grid method is used to produce more reliable and credible 3D images like elevation and contour/isoline maps, I continued to use this method for producing these end products. In general, due to random error distribution, and the use of small number datasets rather than highly densely packed datasets, this has minimal effect on the overall outcomes. Since grid cell maps are the only maps that can be used to accurately produce contour/isoline maps of the results, this centroid point lat-long location error is ignored for the time being. (Personal Notes: All of this is easily correccted in ArcGIS through polygon overlays tools. Square grid cells have an inherent error as well–see my page on hexagonal cell grid mapping. I prefer hexagonal grid mapping, but due to the sqls, time and software limitations, a simpler square cell grid map was overlaid on the data instead. )

The production of these maps began with just data on the medical conditions, but came to incldue other metrics as well coded into the national registers, such as Emergency Visit codes and V-codes. The HICL codes and other coding could be employed as well, but due to time and tehcnological constraints, was employed for population pyramid analyses. but not the grid cell National Population Health study.

Most of these maps do not display of the entire US, just the continental proper. In the first phase of this work, a base mapping projection technique had yet to be developed, so the earliest maps bear a white background with no obvious boundaries. The second phase of this mapping included a filling in of these gaps by converting missing data in true non-zero data, used to porject fake baseline information onto the final map product. This step had to be taken because due to the nature of some of the programming employed, missing data is not acceptible, and so some information has to be placed in all cells defined on the grid map.

This ended up being a very useful step in the analysis. The baseline cells where no data originally existed in the linked zip code database are assigned a very low value which can be unique coded to map as grey background. This production of a grey background to improve the results was later employed in the pure zip code mapping as well.

After trying a number of methods of displaying the information, I favored the use of a false 3D bar or pillar map (z-axis), graphed according to lat and long data, with images produced that rotate 360 degrees. This was due to its ease in programming and simplified, understandable means for displaying the results. These were produced by programming for images with a constant tilt and rotations occuring in increments of 4.2 degrees.

A few examples were produced that ended up with smaller *.MOV files, by applying this method just to a partial rotation, slightly less than 180 degrees, moving back and forth, left-right (swaying).

The majority of the maps are

  • Zip Code N (counts of cases)
  • Zip Code IP (Prevalence)
  • Gridcell N
  • Gridcell IP

By viewing the z-axis values, one can tell an IP map. These usually have “Weight” values of less than 1.0, although exceptionally large N ICDs might have many values near or even above 1.0 due to the nature of the programming.

Grid maps are noticeable at perfect 90 degree angles, like o, 90, 180 and 270. Usually the 90 setting is the best image for identifying these grid cell borders or matrices.

Some squared grid cell or zip code count value maps were produced, to demonstrate the value of squaring baseline results. This is typically used in photogrammetrics and remote sensing imagery to strengthen the topographic differents (accentuating mountain peaks, steepening hills, etc.) It is used here to point out peaks in the data stats. For contagious diseases, this is used to accentuate the nexuses or niduses (nest) of a disease, and can be interpreted along with the contour map usually preceding every ICD data series that was run.

In all, about 500 ICDs/vCodes/eCodes were evaluated using this method. Those capable of demonstrating important interpretation results were chosen. Usually rare to very rare diseases best demonstrate such information. Very common diseases like diabetes, hypertension, hyperlipidemia, tend to produce a very busy map, one with 3D features noticeable and useful for calculations, but not good for demonstrating this methodology with. Besides, for common diseases, a different routine is used to remove the noise of the seemingly endless numbers of cases being mapped.

IP or N?

Generally speaking, some medical events tell us more when N values are mapped, versus prevalence (IP) values. One doesn’t expect a very rare ICD such as bioterrorism (an eCode) to be mapped for all of the events, and then have an IP calculated nationwide. This is because these cases are very rare–few and far between, even on a grid map with merged zip code data, and even when a single case exists, it is important to map it. There are a number of Vcodes and eCodes that have very small counts, but very high social importance. These are therefore more useful and productive for mapping the N values (i.e. see refusal of immunization maps)

IP does provide us with important insights at times , and therefore is included here for review for a number of ICDs.

In terms of other applications of these maps, such as evaluating needs and costs, only the numbers of people matter, not the prevalence. Interpreting this issue from a business angle, a city with tens of thousands of diabetics still has a cost attached, even if its prevalence is the lowest in the country. Therefore, in the long run, N is much more important and useful than IP.

.

. . . . . SORRY, ONLY THE UNEDITED EARLY VERSIONS ARE PUBLISHED UNTIL JULY. . . . . .

Lyme Disease

  • Lyme Disease (National) — http://youtu.be/jJh26LRU2PU. Many of the lyme disease maps are my earliest, relating to a 1998 PSU institutional grant funded program. Some are very basic and hard to follow due to faster image processing speeds.
  • Lyme Disease 2 — New England-New York region — http://youtu.be/lFCQYtqv7rc. Nice focus on the US nidus for Lyme Disease. This disease originated in the area of Sweden but we hear very little about this part of the disease history.
  • Lyme Disease, IP — http://youtu.be/0kVntGbNk2E. A much later version of representation for this data, about 10 generations/renderings into the production of a disease mapping technique.

Rare to Infrequent Diseases

Occupational Diseases

Environmental Exposure Pulmonary Conditions

  • Talcosis/Silicosis (bilayer view) [502] — http://youtu.be/XhPsSijkGSk.
  • Talcosis, Silicosis Lungs (502) — http://youtu.be/tbEegzeqKsI
  • Pneumoconiosis [500-508] (General Review) — http://youtu.be/KP1K9vMD–c
  • Asbestosis [501] — http://youtu.be/jIUOMVRhOs8. A later generation of Asbestosis has developed. Once an occupational disease related to mining, factories and home construction, now a disease related to local economic history. This disease went from being very naturally defined to one defined mostly by human geography and history.
  • Metallic Particle Lung Disease [503] — ; IP — http://youtu.be/EKs3mu3tqaI
  • Stannosis, Beryllosis [503] —
  • Occupational Bronchio-alveolar Diseases [] — http://youtu.be/UveOxuKXBqg. A series of allergy-related diseases with unique occupational causes. Many of these occupations are geographically defined and defined by land use patterns. Others are technologically defined and depend upon the placement of factories more than natural resources.

Moral and Ethical Medical Issues

Genetic or Hereditary

Congenital and/or reduced Lifespan

  • Congenital Lobster Claw Deformity [755.58]– http://youtu.be/PPeiDkhrgkI
  • Horseshoe Kidney — http://youtu.be/PPeiDkhrgkI
  • Louis Bar Syndrome, lt 18 yo [334.8] —
  • Werdung Hoffman Spinal Degeneration, lt 18 [335.0] —
  • Hereditary Choroid Dystrophy, Children [363.5] —
  • Tay Sachs, Children [330.1] —

Newborn/Pediatric Health. The following are sad statements about these important public health issues involving young children.

Childhood, Teen & Young Adult Health

  • Tourette’s Syndrome — http://youtu.be/zXfq1CUW-9w
  • Explosive Social Misconduct Disorder — http://youtu.be/JjsyrcE12fs
  • Preteen/Teen Suicide — http://youtu.be/YEDFvPAU0lU
  • Preteen Homelessness —
  • Teenage Homelessness —
  • Cannabis – ; IP — http://youtu.be/utaqySLgWC4
  • Cholera, all types (001.*, KIDS) —
  • Shigella, all types (004.*, KIDS) —
  • Amoebiasis (006, KIDS) —
  • Other Protozoan (007, KIDS) —
  • Tuberculosis (010-018, KIDS) —
  • Schistosomiasis (all, 120.*, KIDS)
  • Liver Flukes (All, KIDS) —
  • Tapeworm (All, KIDS) —
  • Filaria (All, Kids) —
  • Nutritional Marasmus, less than 13 [261.*] (KIDS) —
  • Low Protein Caloric Uptake, Moderate, lt 13 yo [263.0] —
  • Low Protein Caloric Uptake, Mild, lt 13 yo [263.1] —
  • Low Protein Caloric Uptake, Severe, lt 13 yo [263.2] —
  • Low Protein Caloric Uptake, ALL, lt 13 — [263.*] —
  • Vitamin A Deficiency, lt 13 yo [264] —
  • Various Vitamin/Nutrition Deficiencies, lt 13 yo [265.1 to 266.9] —
  • Scurvy [267]
  • Child with Chronic Disease and Secondary Conditions
    • 5-11 yo
    • 12-17
    • All

Adult Health

Sociocultural Conditions or Diseases

Sociocultural Physical Conditions or Diseases

Socioeconomic/sociocultural related events or medical conditions

Psychological/Psychiatric Health

Poverty and Homelessness

  • Homeless
    • 00-15 yo
    • 16-25 yo
    • 26-35 yo
    • 36-45 yo
    • 46-55 yo
    • 56-65 yo
    • 66+
  • Inadequate Housing [V60.1]
    • 00-12 yo
    • 13-25 yo
    • 26-64 yo
    • 65+
  • Inadequate Material Resources
    • 00-12 yo
    • 13-25
    • 26-64
    • 65+

Poverty/QOL/SES-related Diseases, Behaviors, Conditions, Problems

Infectious Diseases of Foreign Origin capable of In-migration and Persistence

Zoonotic Diseases

Foreign Disease Introduction to the US

Very American born and bred (or named) Diseases

Other Infectious Diseases (not included on any of the above lists)

  • Streptococcal Sore Throat [034.0] —
  • Scarlet Fever [034.1] —
  • Trichomonas — http://youtu.be/0tMhtRnWWws
  • Bacterial Meningitis, children [320.*] —
  • Escherischia coli, Unspecified [008.00] —
  • E. Coli, Enteropathogenic [008.01] —
  • E. coli, Enterotoxigenic [008.02] —
  • E. coli, Enteroinvasive [008.03] —
  • E. coli, Hemorrhagic [008.04] —
  • Clostridium difficile [008.45] —
  • Ancyclostomiasis [126.9]
  • Foot and Mount [087.4] —
  • Rhinosporidosis [117.0] —
  • Sporotrichosis [117.1] —
  • Chromoblastosis [117.2] —
  • Aspergillosis [117.3] —
  • Mycotic Mycetoma [117.4] —
  • Cryptococcosis [117.5] —
  • Allescherosis [117.6] —
  • Phaeohyphomycosis [117.7, 117.8] —
  • Schistosomiasis haematobium [120.0] —
  • Intestinal Schistosomiasis [120.1]
  • Schistosomiasis Japonicum [120.2] —
  • Liver Flukes (General) [121.*] —
  • Feline Liver Flukes [121.0] —
  • Chinese Liver Flukes [121.1, 121.2, 121.5]
  • Farm Liver Flukes [121.3] —
  • Pork Tapeworm [123.0] —
  • Beef Tapeworm (Taenia sagitata) [123.2] —
  • Fish Tapeworm (Diphyllobothriasis) [123.4] —
  • Rat Tapeworm (Hymenolepiasis) [123.6] —
  • Canine Tapeworm [123.8] —
  • Bancroft Filaria [125.0] —
  • Malayan Filaria [125.1] —
  • Onchocerciasis [125.3] —
  • Dipetalonemiasis [125.4] —
  • Mansonella ozzardri [125.6] —
  • Tapeworm (General) (KIDS) —
  • Liver Flukes (General) (KIDS) —
  • Filaria (General) (KIDS) —
  • Meningo Toxoplasmosis [130.0] —
  • HeartLungLiver Toxoplasmosis [130.4,130.6] —
  • Multisystemic Toxoplasmosis [130.8] —
  • Toxoplasmosis (General) (LT 13 yo) —
  • Ainhum (Dactylosis spontanea) [136.0] —
  • Behcets [136.1] —

Other Diseases with Geographic/Topographic Display

Ecodes (Emergency topics)

Vcodes

Other Unique Methods/Topics

Special Training

Miscellaneous

.

.

.

Leave a comment