Skip to 0 minutes and 9 secondsWelcome again to the topic of health facility planning accessibility and GIS. In the first two videos, we took a look at how distributing scarce resources in geographic space plays a role. The second video, we looked into some detail of the accessibility concept. In this video, we get some more insight in commonly used GIS-based measures of accessibility that are used in the geo-health domain. Already mentioned in the previous video is that the link between the accessibility concept and GIS is a very good one. We have seen this diagram before, and we know that in a GIS, we can model the people, the transport, and the opportunity dimensions in a very easy way.
Skip to 0 minutes and 58 secondsCommonly used accessibility measures in the health domain are three-fold. We have simple distance measures. We have cumulative opportunity measures, and we have gravity measures. In reality, there are many different measures, and many of them are more advanced. But in this video, we constrain ourselves to these three because they are widely used. The first are simple distance measures. Here we measure the distance of a point of origin to the nearest health facility, or the reverse. We take the location of a health facility, and we look how many people they are capable of serving. We use a distance threshold, in many cases a straight line distance or Euclidean distance. Network distances are possible.
Skip to 1 minute and 44 secondsThe big advantage of these type of measures is that they have very limited data requirements. They're easy to generate, and there also, the results are easy to interpret. Their big drawback is that these measures are not suited for scenario development or for the generation of planning alternatives, and as such, this type of measures do not really fit in planning. The second group of measures that I would like to present to you shortly is called counter measures. Now, these are quite different in approach because here, what we do is we count the number of opportunities that can be reached within a fixed travel time distance. Let me try to give an example.
Skip to 2 minutes and 29 secondsImagine that you wake up in the morning, and you think of the shops that are not too far from your home. If you increase the distance that you are willing to travel to visit a shop, you will see that more shops become available. If you wish to travel only a very short distance, the number of shops will be limited. The advantage of contour measure is that they have very modest data requirements as well. Also, the results that they produce are easy to interpret, and the skills needed to generate contour analysis are not very advanced. Their big benefit is that they are suitable for scenario development and evaluation of alternative scenarios.
Skip to 3 minutes and 11 secondsThe drawback is that we have to define a distance threshold, and defining a distance or travel time threshold by definition is subjective. And of course, we have to be aware that a different distance threshold will affect the results. So this is quite a drawback of contour measures. Let me try to give you some idea how these contour measures work. Here, we see a very simple example. You see a study area which is subdivided into small zones. We call them accidents. And in this example, we see the number seven standing.
Skip to 3 minutes and 55 secondsNow, if we imagine that in each of these zones, one person is living-- so if I want to put a health facility in the zone that we are looking at, we could see that I can service seven people if I put a health facility in that area. I could service the zone itself and the six zones that it surrounds. If we look to the right upper part of the same image, we see some blue arrows.
Skip to 4 minutes and 26 secondsThe zone with the number four would be able to serve four people. It will be able to serve the person living in the zone itself as well as its three neighbors. The conclusion is that the zone that can service seven people has a higher level of accessibility than the one serving only four people. If we extend this to a larger area-- and here we see part of a city-- then we could see that some areas that are more centrally located have a high service capacity, whereas areas which are located in the periphery of this part of the city have a lower service capacity. So this is the way in which contour measures work.
Skip to 5 minutes and 10 secondsThe third type of measure that we would like to discuss here is the gravity measure. The gravity measures are based upon Newton's law of gravity, and they are an index of the potential of interaction between locations depending on their size and the degree of separation. The idea is that the larger the distance between two masses is, the lower the potential of interaction. At the same time, the larger the size of the two bodies that interact is, the larger the interaction will be. So the interaction potential is a measure of geographical separation on the one hand and size on the other hand.
Skip to 5 minutes and 54 secondsImportant also here is that the level of interaction is calculated between one location respective to all other locations in the given study area. This makes that this type of measure is very suitable for the development of scenarios and their evaluation. They are very widely used as well. They also have a drawback. Because they are a relative measure, the interpretation of the results is not always easy. Another difficulty is that depending on the degree of separation, the probability of interaction declines. It is always difficult to estimate how much this decline will be, and normally, we need additional data in able to have a proper estimation of what we call this decay of interaction.
Skip to 6 minutes and 45 secondsLet's look at this particular example of a potential model. What we see here are a number of different locations, A, B, C, D, and E. We also see the distance between each of these two location pairs. If we look at the lower graph, we can see that the distance from A to A is zero. The distance from A to B is eight, from A to C, four, from A to D, nine, and from A to E is 15. When we subdivide the total distance by the number of opportunities, we get a value of 7.2. The interaction potential of location A is 7.2.
Skip to 7 minutes and 34 secondsIf we do this for all of the different locations and we fill that table with factors, then we can see that location C is actually the one with the highest potential for interaction. The reason is that the total difference from C to all the other locations is the lowest. In simple words, location C is the most accessible location in these origin pairs. We can make the potential model a bit more complicated by also considering their size. And again, we see a sample sheet in front of us. In this case, we do not only consider the distance between the different locations, but we also consider the size of the locations.
Skip to 8 minutes and 23 secondsAnd we can see both the table of distance and of size on the upper right-hand of this slide. We move down to the lower table, and here, we try to fiddle the table with a combination of the size of a given origin with the size of a given destination. And we include the distance, so it means we multiply the size of the two interacting objects, and we divide it by the distance that separates the two. If we do this for all the different locations, the last column shows their interaction potential, now considering degree of geographical separation as well as their size. And if we look again, we can see that the location with the highest interaction potential is C.
Skip to 9 minutes and 17 secondsSo also, if we consider size location, C is the most accessible location in this example.
Skip to 9 minutes and 27 secondsWhat did we discuss in this short video? Well, first of all, it is important to acknowledge that GIS is very suited for doing accessibility analysis. We've also tried to give a very, very brief overview of some of the most commonly used measures of GIS and accessibility in health. Important to conclude with is that there is not one single best approach to measuring accessibility. Everything depends on the purpose that you want to attain and on the situation in which you're working. It depends very much on the data requirements, the data availability, the need for ease of interpretation or computational complexity. In some cases, we have to acknowledge that accessibility analysis alone is not enough.
Skip to 10 minutes and 20 secondsWe need to complement this with qualitative evaluations, maybe by involving communities and asking their opinions. Nevertheless, the accessibility concept is a very important concept that can support planning, and this is the topic of the next video, actually health facility planning and accessibility analysis.
Common accessibility measures
In a previous article we introduced the accessibility concept. Here, we shortly introduce the three most commonly used measures of accessibility analysis in the Geo-Health domain: (i) simple distance measures, (ii) cumulative opportunity measures, and (iii) gravity measures.
Simple distance measures The simplest distance measures operationalize accessibility as the straight line distance between two locations in geographic space. This type of measure is often used in situations where standards exist in terms of maximum travel time or distance to a health facility (e.g. every person must be able to reach a facility within 30 minutes travel time). This type of approach is very commonly used and operationalized using standard GIS functionality through the generation of service areas around opportunities using functions such as buffers or Thiessen polygons. An advantage of these measures is that data requirements are modest and results are easy to interpret. A major disadvantage is that this type of measure is not suited for the development of scenarios for health facility planning.
Cumulative opportunity measures
Cumulative opportunity measures count the number of opportunities - or number of potential customers - that can be reached within a given travel time or distance threshold. Accessibility increases if more opportunities can be reached within a given travel time or distance. Their main advantage is that data requirements are modest and that results are easy to interpret. Another important advantage is that alternative planning intervention scenarios can easily be generated and evaluated. An important drawback is that the choice of a cut-off travel distance or time is subjective (especially if travel behaviour of patients is not known). Another weak point is the deterministic assumption that all customers will always visit the nearest opportunity.
Briefly, gravity measures - analogous to Newton’s Law of Universal Gravitation - stipulate that two places will interact with each other in proportion to the product of their size and inversely according to some function of the distance between them. In short, the accessibility of a particular location is a function of its relative proximity to all alternative destinations in a given spatial system. Generally speaking, the more accessible a location is, the higher the potential of spatial interaction with surrounding locations becomes. Gravity measures are suitable for developing and evaluating alternative intervention scenarios. An important advantage of gravity measures is that they are not deterministic - patients are not all assumed to attend the nearest health facility - and that their data requirements are undemanding. Compared to measures of cumulative opportunity, however, the calculated accessibility indicator is harder to interpret. Gravity measure also require the estimation of a distance decay function to describe the friction of distance (normally this is based upon a travel survey).
For a more detailed explanation about this topic see the attached file below.
References: Handy S.L. and Niemeier, D.A. (1996). Measuring accessibility: an exploration of issues and alternatives. Environment and Planning A 1977, vol. 26. pp. 1175-1194
Higgs, G. (2004). A literature review of the use for GIS-based measures of access to health services. Health Services and Outcome Research Methodology5: 119-139.
© University of Twente