Comparing alternative methods to measuring pedestrian access to community pharmacies

The aim of this paper is to compare several methods for measuring geographical accessibility to community pharmacies in the Lisbon Metropolitan Area (LMA). Twelve measures of pedestrian distance between spatial units and the closest community pharmacy were computed based on the combination of 4 parameters: type of distance, location, centroid definition, and level of spatial unit. For this, the Google Maps Application Programming Interface was used for calculating network pedestrian distances, using a list of 801 community pharmacies and population data from the Census 2011. Correlations between every method were performed, and the variations of the estimated number of inhabitants served at an 800-m distance were analyzed. Local errors were assessed comparing every combination to the most accurate one. The results show that the number of people served ranges from 70 to 89% of the total population, depending upon the method used. The use of pedestrian network distances decreases by more than 10 points the population coverage, compared to crow-fly distances. The finest parameters of population assignment are more inclusive than coarser ones. This research demonstrates the influence of several measurement methods on coverage estimations. Empirical evidence indicates that both measurement and policies should be called into question in order to improve actual coverage