From rdf4@cdc.gov Mon Jul 9 15:23:23 2001 Received: from mxu2.u.washington.edu (mxu2.u.washington.edu [140.142.32.9]) by lists.u.washington.edu (8.11.2+UW01.01/8.11.2+UW01.04) with ESMTP id f69MNJ0119594 for ; Mon, 9 Jul 2001 15:23:19 -0700 Received: from mcdc-us-smtp3.cdc.gov (mcdc-us-smtp3.cdc.gov [198.246.97.19]) by mxu2.u.washington.edu (8.11.2+UW01.01/8.11.2+UW01.04) with ESMTP id f69MNIK31108 for ; Mon, 9 Jul 2001 15:23:18 -0700 Received: from mcdc-us-tis.cdc.gov ([158.111.6.74]) by mcdc-us-smtp3.cdc.gov with SMTP (Microsoft Exchange Internet Mail Service Version 5.5.2653.13) id 3K1DN761; Mon, 9 Jul 2001 18:24:17 -0400 Received: by mcdc-us-tis.cdc.gov with Internet Mail Service (5.5.2653.19) id <3KFH7DK2>; Mon, 9 Jul 2001 18:23:08 -0400 Message-ID: From: "Friedman, Roger" To: "'waphgis@u.washington.edu'" Subject: RE: Zip Codes, ZCTA and areal interpolation. Date: Mon, 9 Jul 2001 18:23:05 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" I believe there have been some initiatives to push geocoding back to the source. Certainly with the web-based geocoding tools available and the phone-to-location capability existing within EMS operations, it ought to be easier to capture at least some geocoding information earlier in the process. As far as getting the information we need, that would require a lot more intelligence than we exhibit now. For example, I saw a study that used geographic methods on a food-related outbreak where the only geographic variable available was the location of treatment. The methods "worked," but even geocoding to patient address would not have improved the study's value. Certainly there are a large number of public health threats where we would like to know everywhere the patient has been for 72 hours prior to treatment; these are the same threats where confirmatory lab work is not being done. (Perhaps we should be mining the cell phone records of our study participants. Then we would be suffer from the same data reduction problems the population ecologists are having.) It seems that the tried and true methodology of epidemiology is to do now what you can do now, and what we can do now with GIS is more than what we are doing. But that doesn't necessarily mean we should increase our requirements for geographic data. Although I think GIS is exceedingly useful for health data presentation, I think ultimately we'll get more information useful for disease control from studies of the genetics and protein dynamics of disease. It may be more important that we make sure we are grabbing DNA samples or microarray readings. Off the soapbox, Roger Friedman -----Original Message----- From: Richard Hoskins [mailto:healthmaps@home.com] Sent: Monday, July 09, 2001 5:41 PM To: waphgis@u.washington.edu Subject: RE: Zip Codes, ZCTA and areal interpolation. I think we need to be very activist concerning geocoding issues - just because an agency says they won't release data below Zip Code doesn't mean we should accept that reply if we have serious public health work to do. If we need larger scale data, then we should get it - we need to push our agency management and use the Institutional Review Board process. I suggest that "don't have the option" is not the case at all - although getting another option may not be all that convenient. Public health people should demand, and I mean that in the most civil and with the most non-confrontational attitude possible, but in the end public health practitioners must get access to the data they need. All public health GIS folks need to deal with this issue up front. Accept nothing less than what gets the job done. Zip Codes or ZCTA's are not a good spatial entity for public health analysis. Changing boundaries, inhomogeneous mix of population, small scale relative to importance to the public health intervention planning needed; these are all issues with Zip Codes. Many states collect data at the Zip Code level only for some important data (in my state, hospital discharge data) but it is practically usefulness for effective assessment much less designing an intervention. We GIS people have not gotten the message across that Zip Codes are for delivering the mail, not for delivering public health services. If health record sources, in fact, have the address level data, then we should push them to adopt policies that allow access to this data when it is the differences between an effective intervention and one that wastes public money and does not serve the people that need it. Richard Hoskins -----Original Message----- X-Sybari-Space: 00000000 00000000 00000000 00000000 From: WAPHGIS-owner@u.washington.edu [mailto:WAPHGIS-owner@u.washington.edu]On Behalf Of Susan Clemens Sent: Monday, July 09, 2001 1:04 PM To: waphgis@u.washington.edu Subject: RE: Zip Codes, ZCTA and areal interpolation. > ---------- > From: Susan Clemens[SMTP:S-CLEMENS@WILDMAIL.COM] > Sent: Monday, July 09, 2001 1:03:30 PM > To: waphgis@u.washington.edu > Subject: RE: Zip Codes, ZCTA and areal interpolation. > Auto forwarded by a Rule > I have been skeptical of zip codes and/or ZCTA for some time now. Unfortunately, I am usually analyzing data collected by another agency so don't have the option of requesting more specific geographic information. Often a zip code is all that is collected or all that can be released due to patient confidentiality issues. I agree with the observation that zip codes are not consistent through time. As I understand it, perimeter changes are the norm when designating new zip codes. Also, zip + 4 numbers change more frequently than zip code perimeters and often have more to do with postal carrier staffing and routes vs population. Lastly, the ZCTA for the 1990 Census were developed using a different methodology than the 2000 ZCTA. Even if you are lucky enough to be in an area with no new zip codes you cannot compare 1990 and 2000 ZCTA data. So, where does this leave us? Is commercial zip code level data any more reliable that the ZCTA? Is any zip code level analysis for health valid given the constraints mentioned? Susan Clemens Humboldt Co Dept of Public Health ---- Begin Original Message ---- From: fpb01@health.state.ny.us Sent: Thu, 5 Jul 2001 15:44:38 -0400 To: waphgis@u.washington.edu Subject: RE: Zip Codes, ZCTA and areal interpolation. I believe that ZCTAs were created for the exact purposes given in Richard's original posting: to provide an easier way to deal with the mass of data that is geocoded to zip code only, and to provide a standard (and free) base map on which to map this data. ZCTAs are probably good enough for many purposes, such as marketing applications. They should also be good enough for certain kinds of proportional analysis. For example, you can probably safely compare a variable such as % smoking from a zip code-based survey with % completed high school from ZCTA. When it comes to using ZCTAs for calcuating incidence and mortality rates, though, I am pessimistic. Frank Boscoe NYS Department of Health "Richard Hoskins" @u.washington.edu on 07/05/2001 03:12:24 PM Please respond to waphgis@u.washington.edu Sent by:  WAPHGIS-owner@u.washington.edu To:   cc: Subject:  RE: Zip Codes, ZCTA and areal interpolation. Frank:  What was the purpose of the census bureau's developing the ZCTAs? Dick H -----Original Message----- From: WAPHGIS-owner@u.washington.edu [mailto:WAPHGIS-owner@u.washington.edu]On Behalf Of fpb01@health.state.ny.us Sent: Thursday, July 05, 2001 12:00 PM To: waphgis@u.washington.edu Subject: Re: Zip Codes, ZCTA and areal interpolation. I believe that conversion from Zip Codes to ZCTAs is, unfortunately, more than just an areal interpolation problem. The two entities are not the same thing - they differ along the edges. On a road that forms the boundary of a zip code, both sides of the road may be in the same Zip Code. This is reasonable and facilitates the delivery of the mail. But the two sides of the road will be in different census blocks and hence different ZCTAs (since ZCTAs are built from census blocks).  In heavily populated areas, the distinction is almost negligible, but not in rural areas. In fact, the census produced a graph which shows that zip code-ZCTA agreement is almost perfect in zip codes with more than 15,000 addresses, but under 80% where there were fewer than 250 addresses. In terms of health mapping, if a nursing home or hospital or other special type of address falls on a zip boundary, its entire caseload could be assigned to the wrong ZCTA, resulting in substantial error even in a heavily populated area. Frank Boscoe NYS Department of Health ---- End Original Message ---- Care2 make the world greener! http://www.care2.com - Get your Free e-mail account that helps save Wildlife! .