New Mexico Health Policy Commission


The New Mexico Health Policy Commission (NM HPC) and the Division of Government Research (DGR) at the University of New Mexico (UNM) have worked on developing a gravity model useful in measuring geographic access to health care facilities and providers. Preliminary tests and initial applications of this gravity model (based on ZIP Code geography ) have produced very promising results. A description and examples of this gravity model will be made available here as further refinements are made and when the NM HPC authorizes this material for public release.

This gravity model is currently operational in SAS (Statistical Analysis System) and as a SAS Macro. The results (accessibility measures by ZIP Code ) are converted from a SAS dataset to an ESRI ARC/INFO export file (using SAS Macros developed by DGR) or a dBase (.dbf) file. This ARC/INFO export file or .dbf is imported to ESRI's ArcView where maps (views and layouts) of the results are produced. Examples of these maps will be presented when the NM HPC approves their release. Please DO NOT contact the NM HPC for more information. DGR will respond to any questions about this technique..

The Division of Government (DGR) and the New Mexico Health Policy Commission (NM HPC) won the poster contest for best analytical content at the ESRI Southwest User Group Conference (SWUG) held in Taos, New Mexico October 2002. This poster entitled "Measuring Geographic Access to Primary Care Physicians in New Mexico" is available for download as a pdf. Click here to obtain this poster in either 11x17 inches or 34x44 inches formats.

Some example maps and further description of DGR's gravity model are available in a recent PowerPoint presentation.

Another implementation of this gravity model has been developed using Microsoft's Excel. Special scripts (using Avenue and Microsoft DDE) have been developed to link the results to ESRI's ArcView . Future work may include application development with ESRI's ArcObjects using Microsoft's Visual Basic .

This information is being provided as a courtesy to other researchers that are conducting similar studies. It is hoped that the work presented here will be useful to everyone currently considering developing gravity models and other techniques for measuring geographic access to health care. Please remember that these results are preliminary and more work needs to be done refinining this gravity model. Also, the preliminary maps should be viewed with caution because the data is sometimes incomplete and the the analytical methods are under development.

The Issues:

Older discussion - See SWUG Poster & PowerPoint presentation(above) for recent description!

Issues related to the study of geographic access to health care have been the focus of many academic and applied research studies during the past 30 years. Most of this research has been undertaken in an effort to develop better tools for government and corporate decision makers whose task it is to develop better management strategies, planning guidelines, or policies. The rationale behind these efforts is that better information and information analysis techniques should result in better decisions.

Most studies of geographic access to health care can be broadly categorized based on their focus upon either need or demand issues (Connor, Kralewski and Hillson 1994). Need-based studies are based on the provision of health care to a population regardless of market conditions or consumer ability to pay. Demand-based studies take into consideration the influence of economic factors such as price, insurance coverage, and medical practice arrangements.

Another useful classification of studies of medical care delivery that are concerned with spatial dimensions have used the terminology "revealed accessibility" and "potential accessibility" in an attempt to define the different focus of previous studies (Joseph and Phillips 1984, Phillips 1990, Thouez et. al. 1988). Revealed accessibility studies primarily focus upon the patterns of utilization as being the result of associated behavioral processes. Potential accessibility studies tend to emphasize the geographic patterns and aggregate supply of medical care resources as a research focus.

The limitation in conducting studies based on either of these approaches has traditionally been the availability of data. In general, it is easier to acquire the data to perform a need-based and potential accessibility study than those that are demand-based and focused on revealed accessibility. Also, it is generally more difficult to explain economic and behavioral patterns of consumer/patient decisions (demand and revealed accessibility studies) than to simply measure the geographic distribution of facilities and consumer/patients (need and potential accessibility studies).

Depending on the availability and quality of data available, previous research can also be classified somewhere from the simply descriptive (mostly concerned with measurement) to the theoretical (design and application of models). Many of the policy oriented government sponsored studies are concerned with measuring provider availability and defining health service areas (Lee 1991, Makuc et. al. 1985 and 1991). The academic and industry sponsored studies tend to focus more on developing models that describe the provision of health care services and even predict the location and allocation of these services with respect to alternative scenarios (Hodgson 1984). They have also focused on predicting consumer behavior and the rigorous definitions of market areas for various types of health care services (Cowper and Kushman 1987, Garnick, et. al. 1987).

Because of current data limitations, DGR has concentrated on developing and employing a need-based and potential accessibility measure of geographic access to health care. In essence, this gravity model is only an elaborate measurement technique. It is designed to measure the supply of medical resources in relation to the population being served.

Traditional measures of the availability of health care resources have not been very good at taking into consideration the supply of resources in a community, the supply in neighboring communities, and the distance and ease of travel among them (Klienm an and Makuc 1983). These traditional manpower measures (GMENAC 1980, Wing and Reynolds 1988) are basically ratios of the number of providers or facilities by geopolitical unit (county, census tract/block, or zipcode). They fail to account for consumers travelling between adjacent or relatively close geopolitical units to acquire health care services. They can also result in overestimation in some areas and underestimation in others.

The gravity model technique has a long history of application in the economic and social sciences (Carrothers 1956, Haynes and Fotheringham 1984, Huff 1963, Reilly 1929). It has also been used in many health care oriented studies (Gesler 1986, McGlashan and Blunden 1983, Guptill 1975, Pyle 1979, Shannon and Dever 1974). The reason gravity models have proven useful is that they allow for the interaction between the phenomena being studied such as that between shoppers and retail stores or patients and health care providers to be measured in relation to distance or travel time.

Simply stated, interaction is assumed to decline with increasing distance. The result of declining interaction is termed distance decay and this effect is termed friction of distance It is usually approximated by a negative exponential function for the distance measure in traditional gravity models. When used to measure the accessibility of health care a gravity model is usually an ideal choice because interaction or utilization of facilities, services, or providers by consumers/patients drops off with increasing distance (distance decay).

Another useful element of a gravity model is that they allow interaction to be measured in a cumulative (or weighted) fashion regardless of arbitrary borders or boundaries. For instance, consumers/patients routinely travel across arbitrary geopolitical units such as counties, census tracts, or zipcodes to shop or receive health care. In measuring the flow of consumers/patients to a given store or health care provider a gravity model is useful because it derives a cumulative number that considers all the potential and distant consumer/patients.

DGR’s Gravity Model:

The gravity model employed by DGR has a sound foundation in previous research (Knox 1978, Joseph and Bantock 1982, Khan 1992) focused on developing a measure or index of potential accessibility of health care providers. However, DGR is using a more basic variation. This is primarily because not all the data necessary to address concerns such as potential demand and relative mobility of consumer/patients, additional concerns of the previous studies, is available.

The primary goal of DGR’s gravity model is simply measurement. A need based measure of the potential accessibility of health care providers and facilities had to be developed for the HPC data (Primary Care, Medical Specialties, and Behavi oral Health Care Services Survey) that was available. This measure had to be better than traditional manpower ratios based on areas such as counties or zipcodes. It had to take into consideration interaction across boundaries and also to realistically model this interaction as a distance decay function. Because it attempts to quantify interaction it can also be said that the secondary goal of this gravity model is to take into consideration some realistic assumptions about consumer/patient behavior as part of this measurement.

The gravity model employed by DGR derives a unit of measure for potential accessibility that correspond with the standards established for New Mexico (NM GADS Workgroup 1997). The unit of analysis is zipcodes. In order to simplify the analysis, the location of a principal place (usually a settlement containing a post office) is used to define a concentration of population or providers at a singular point in space.

DGR’s gravity model can be expressed by the following equation where PA is a measure of potential accessibility using zipcodes represented by city points at point j; pop is population; prov is the number of providers being measured (hospital beds or dentists or primary care, etc.); and d is distance between points i and j:

This gravity model in its simplest form is nothing more than an elaborate ratio measure of the weighted (summation across zipcodes) population to weighted (summation across zipcodes) number of providers for each zipcode. It becomes slightly more complicated when you notice that it is also a compound gravity model consisting of a gravity model for population as the numerator and a gravity model for providers as the denominator. Further, the weighting or summation across zipcodes decreases proportionally to distance.The weighting is done in three distance zones: up to 35 miles the weighting equals one; between 35 and 100 miles the weighting decreases proportional to the square of the distance; beyond 100 miles the weighting equals ze ro. The constant of (35)2 the distance decay portion of the weighting function, it compensates for the 35 mile "friction free" area. The 35 mile distance was chosen to correspond approximately to the 45 minute travel time referred to in the standard (NM GADS Workgroup, 1997).

All distances used in this compound gravity model were measured in a straight line between points representing the principal place (post office location) in each zipcode. The population for each zipcode is based on the number of licensed drivers in tha t zipcode. The Census Bureau 1997 New Mexico statewide population estimate was used to adjust the driver counts to more realistically estimate population by zipcode.

Current Applications:

DGR has applied this compound gravity model to several provider based data sets derived from recent HPC data. SAS and Excel were used to build the model and perform the analyses. ArcView was used to create the maps. The several applications have been:

These several gravity model applications have demonstrated the utility of this measure of potential accessibility. However, some refinements and improvements can be made for the future.

References and Bibliography:

Carrothers, G. A. P. "An Historical Review of the Gravity and Potential Models of Human Interaction." J Amer. Inst. 0f Planners 2 (1956): 94-102.

Connor, R. A., J. E. Kralewski, and S.D. Hillson. "Measuring Geographic Access to Health Care in Rural Areas." Medical Care Review 51:3 (Fall 1994): 337-377.

Cowper, P.A., and J.E. Kushman. "A Spatial Analysis of Primary Health Care Markets in Rural Areas." American Journal of Agricultural Economics 69 (1987): 613-625.

Garnick, D.W., H.S. Luft, J.C. Robinson, and J. Terault. "Appropriate Measures of Hospital Market Areas." Health Services Research 22, no. 1 (1987): 69-89.

Gesler, W. "The use of Spatial Analysis in Medical Geography: A Review." Social Science and Medicine Vol. 23 No. 10 (1986): 963-973.

GMENAC (Graduate Medical Education National Advisory Committee), Summary Report. DHHS Pub. No. (HRA)81-651. Washington, DC: DHHS, 1980.

Guptill, S. C. "The Spatial Availability of Physicians." Proc. Ass. Of Am. Geogr. Vol. 7, (1975): 80-84.

Haynes, K. E., and A. S. Fotheringham Gravity and Spatial Interaction Models. Sage Publications, Beverly Hills, Calif. (1984).

Hodgson, M. J. "Alternative Approaches to Hierarchical Location-Allocation Systems." Geogr Anal. Vol. 16 (1984): 275-281.

Huff, D. L. "A Probability Analysis of Shopping Center Trading Areas." Land Economics 53 (1963): 81-90.

Joseph, A. E., and P. R. Bantock. "Measuring Potential Physical Accessibility to General Practioners in Rural Areas." Social Science and Medicine 16 (1982): 85-90.

Joseph, A.E., and D.R. Phillips. Accessibility and Utilization: Geographical Perspectives on Health Care Delivery. London: Harper and Row, 1984.

Khan, A. A. "An Integrated Approach to Measuring Potential Spatial Access to Health Care Services." Socio-Econ. Plann. Sci. Vol. 26, No. 4. (1992): 275-287.

Kleinman, J.C., and D. Makuc. "Travel for Ambulatory Medical Care." Medical Care 21, no. 5 (May 1983): 543-557.

Knox, P. L. "The Intraurban Ecology of Primary Medical Care: Patterns of Accessibility and Their Policy Implications." Environment and Planning A Vol. 10, (1978): 415-435.

Lee, R.C. "Current Approaches to Shortage Area Designations." Journal of Rural Health 7, no. 4 (1991): 437-450.

NM GADS Workgroup. GADS Workgroup Recommendations on the Geographic Access Data System (GADS) and New Mexico Specific Health Access Standards, An Interim Report. HIS Advisory Committee, NM HPC, Santa Fe, NM (1997).

Makuc, D. M., B. Haglund, D. D. Ingram, J. C. Kleinman, and J. J. Feldman. "The use of Health Service Areas for Measuring Provider Availability." Journal of Rural Health 7, no. 4(1991): 347-356.

Makuc, D., J.C. Kleinman, and B. Pierre. "Service Areas for Ambulatory Medical Care." Health Services Research 20, no. 1 (April 1985): 1-18.

McGlashan, N. D. And J. R. Bluden (eds.) Geographical Aspects of Heath. Academic Press London, 1983.

Phillips, D. R. Health and Health Care in the Third World. Longman, Harlow, Essex (1990).

Pyle, G. F. Applied Medical Geography. Winston, Washington, D.C. (1979)

Reilly, W. J. Methods for the Studying of Retail Relationships. University of Texas, Monograph No. 4, Austin, Tex. 1929.

Shannon, G. W. and G. E A. Dever. Health Care Delivery: Spatial Perspectives. McGraw-Hill. New York. 1974.

Thouez, J. P., P. Bodson, and A. E. Joseph. "Some Methods for Measuring the Geographic Accessibility of Medical Services in Rural Regions" Medical Care 26, no. 1 (January 1988): 34-44

Wing, P., and C. Reynolds. "The Availability of Physician Services: A Geographic Analysis." Health Services Research 23, no, 5 (December 1988): 649-667.


  • For more information:

         New Mexico Health Policy Commission
         Health Information System
         2055 S. Pacheco  Suite 200 
         Santa Fe, New Mexico  87505
         Phone: (505) 424-3200 
         FAX: (505) 424-3222 
         Division of Government Research
         University of New Mexico 
         Onate Hall Suite 116  MSC06 3510 
         Albuquerque, New Mexico  87131
         Phone: (505) 277-3305 
         FAX: (505) 277-7066 
         Email: e-mail 

    Mail Send Mail to DGR  

    UNM DGR's Home Page     UNM HPC's Home Page

    UNM UNM's Home Page

    NM NM Government Information Homepage

    Created by Division of Government Research
    University of New Mexico

    Last modified on 10/26/2004 by Larry Spear e-mail