Geographic Access Gravity Model 1. Statement of Problem (Measurement) 2. The Theory and Method (Potential Accessibility) 3. Applications (Preliminary Maps) 4. Future Directions (Data & Model Improvements) 5. Feedback from Others (Discussion) 1
Jan 21, 2016
Geographic Access Gravity Model
1. Statement of Problem (Measurement)
2. The Theory and Method (Potential Accessibility)
3. Applications (Preliminary Maps)
4. Future Directions (Data & Model Improvements)
5. Feedback from Others (Discussion)
1
The Problem(s) - Measurement
• How to MEASURE geographic access to health care providers and facilities?
• The Solution - Develop a reliable method to MEASURE (and compare) the distribution of facilities/providers and the population.
• Reliable MEASUREMENT requires a geographic framework in which to collect and organize observations.
2
The Problem(s)
• Reliable MEASUREMENT requires a common scale that allows for comparison of values.
• Reliable MEASUREMENT requires a method to handle arbitrary boundaries imposed by a data collection geographic framework.
3
The Geographic Framework - ZIP Codes
• Health care data (patient, provider, facility, etc.) have many geographic (locational) components.
• Some geographic components are; geographic coordinates, county, census block and tract, etc.).
• An address with a ZIP Code is a widely used geographic component (our choice).
4
NM ZIP Codes
• US Postal Service delivery areas.• Some ZIP Codes do not have rural
delivery and mail is picked up at the Post Office.
• There can be multiple ZIP Codes per post office (delivery and no delivery).
• DGR has prepared a ZIP Code base map for NM with 400 ZIP Codes for mapping HPC Data (new ZIP Codes are being added - Currently 403).
• We added (estimated) boundaries for 125 ZIP Codes (USPS review).
5
DGR’s ZIP Code Maps
1999 and 2002
Map of ZIP Codes Added and Estimated
A Common Measurement Scale
• Service Capacity Standards (traditional measure - Fed. and State guidelines).
• Ratio of provider or facilities per population.
• Can be expressed as either:
• One M.D. per 1,500 persons (Prov./ Pop.)
• 1,500 persons per M.D. (POP. / Prov.) **What we use.
8
The Boundary Problem
• Traditional measures (service capacity standards) NOT very good:
• Does NOT consider that people move among communities or political/data collection units to obtain medical services.
• Nobody pays attention to what ZIP Code their doctor is in.
• Does NOT consider distance (close or far).
9
Map of Dentists per ZIP Code
The Theory
• Spatial Interaction - The closer together phenomena are the easier it is for interaction to take place.
• Spatial Interaction - Assumed to decline with increasing distance.
• Distance Decay - The result of declining interaction - termed friction of distance.
11
The Method (Spatial Analysis)
• Gravity Models have been used in economics and social sciences since William Reily (Univ. of Texas) proposed the idea in 1929.
• Gravity Models have been traditionally used in retail studies, but recently in health care.
• Gravity Models allow for the measurement of spatial interaction as a function of distance.
12
Original (Retail) Gravity Model
ij
jiij
dPPI b ijjiij dPPI b
or
ijI Interaction between two areas i and j
ji PP and Population of each area
ijd Distance between areas
b Distance exponent - the higher the greater the friction of distance
** Market area definition (polygons)13
Potential AccessibilityDGR’s Gravity Model
n
i
iij
n
i
iij
j
provdf
popdfPA
1
1
Code for ZIPity Accessibil PotentialPAj
Code ZIPa of Population ipop
Code in ZIP FacilitiesProviders/ ofNumber iprov
100 d if 0
100 d 35 if 235 d if 1
ddij 14
Applications
• Still doing initial development and testing.
• Preliminary survey data used to test gravity model.
• Test maps for:– Hospital Beds– Primary Care Physicians– Registered Nurses– General Dentists
15
Applications
• New data available shortly from licensing boards.
• Demographic data being evaluated.
• Risk factor (population segments at risk for receiving inadequate care) applications based on demographic data are being considered.
16
Preliminary (Test) Gravity Model Maps
Future Directions
• Use new data from licensing boards on a quarterly basis.
• Enhance gravity model for demographic and risk factor analyses.
• Train HPC staff to use the Excel and SAS based versions along with ArcView 3.x and Avenue scripts.
• New version using ArcGIS (ArcMap) and ArcObjects (VB/A).
19
Future Directions
• Address the “Edge Problem”:– Peripheral area data (Arizona,
Colorado, Texas and Mexico ??)
• Investigate using travel time instead of straight line distance (NM roads Arc/Info network coverage)
• Review recent literature.• Prepare a presentation and
publication.20
Questions and Open Discussion
21
Internal Web Page
wwwdgr.unm.edu/hpc/hpc_grav.html
MORE DETAIL - Up to date