FCA-metrics

[TomErik]

Hey guys, I’m looking to compare two different building agglomerations to a facility-type in their spatial access. I’ve read some litterature on the 2-step floating catchment metric, and different variations of this. Would anyone care to explain to me the primary difference between this method and using Network Analyst?:) I understand that it uses scores as opposed to travel time, and that it may use networks as a way to measure distance. Could these scores be compared across analyses, say if I ran the analyses twice on different agglomerations?

Best,

Tom Erik

[CarstenSchuermann]

Well, Tom, you can´t directly compare the FCA metrics with the Network Analyst. The Network Analyst provides tools that can support you in the implementation of the FCA metrics. Both steps of the FCA approach rely on the generation of drive time areas (catchment areas). In step 1, you create drive times from a specific healthcare provider (for example, from a hospital or medical office). In step 2, you create drive times from the places of residence (such as resident houses or street blocks). The network analyst can support you in both steps: as part of the network analyst´s  toolbox, the Make Service Area Layer tool generates drive time areas and stores them as polygons. In this tool, you can specify the travel time threshold (such as 15 min, 30 min or 1 hour) and of course it needs the location of healthcare centers or resident areas respectively as input layers. In order to implement the complete FCA approach, you would then need to develop a model or script in ArcGIS where you include the network analyst tool.

Having said that, one more remark on the FCA methodology in healtcare: The FCA approach assumes that people have a freedom of choice as to which healthcare facility (hospital, general practitioner etc.) they visit. Depending on the national health care systems, this is not the case in all countries. In many countries of the world, there is a fixed assignment of places of residence to certain healthcare facilities (by law or by health care regulations). In these countries, using the FCA approach would not properly reflect reality. My advice, therefore, is to examine whether the FCA approach is at all applicable in your example, before you further dig into it and think about how to implement it in network analyst.

[TomErik]

Hey, thanks for the reply. So in cases of acute medical treatment where pasients are transported without a choice, FCA is not applicable? 

[CarstenSchuermann]

I am not saying it is not applicable. Whether the FCA approach is sutiable in your case, depends on several factors, such as

(1) what is your (research) question/problem you would like to answer? If you just want to know a service area of a hospital? Or do you want to match demand for hospitals/doctors with supply? etc.

(2) you need to take into consideration the national health regulations. For example, many countries defined performance indicators like "maximum travel time/distance" to next hospital. These parameters should be accounted for in your analyses. And yes, in some countries patients don´t have a free choice to select the general practitioner or dentist or even the hospital (in other words, the match/assignment between supply and demand has already been decided for by state-government when the government already defined some kind of 'health care zones' - sometimes the definition of such zones already accounted for accessibility, but often such zones where defined based upon general administrative units or other factors).

(3) you may also differentiate between emergency situations (i.e. a patient is transported by rescue services; in this case usually patients are not be asked where to go) and 'normal' situations where patients drive themselves to hospitals/doctors.

(4) you may even also look into the type of medical treatment. Usually, not all health facilities are capable for all kind of treatments, so specialized treatments may only be offered in selected facilities. In such cases, a purely drive time area approach may also not adequately reflect the problem statement.

(5) as part of their health policies, many countries/regions defined also indicators like "number of hospital beds per inhabitant", which at a glance seem to be an easy indicator. But as the actual corona pandemic shows, you also would need to take into a account a temporal dimension (number of available hospital beds per inhabitant at a given time - this usually is often not accounted for in the regulations).

You should think about these (and probably many more) research questions and answer them before you assess whether the FCA approach is suitable or not (and probably how it needs to be modified to best match your research question), and how and where the network analyst can be used to answer your questions.