A simple cost-effective method for random placement of research points

We propose a method using a combination of ArcView and a GPS to accurately and effectively create and locate random sample points over large research areas. We also present a preliminary public health case study which used this method to locate sample sites in a rural area of Limpopo Province, South Africa. Both health and vegetation surveys were carried out at these sites. Although similar techniques have been used in environmental science research we are unaware of the use of combined GIS-GPS technologies for sample determination, nor do we know of use of this approach in health sciences research. Introduction GIS and GPS technology have been used in environmental research for many years, but the use of geo-spatial techniques has only recently begun in medical fields such as public health and epidemiology. Encouragingly, health researchers are beginning to recognize this powerful tool, particularly in mapping the spread of infectious disease and predicting future outbreaks based on spatial characteristics (Pope et al. 1992, Malone et al. 2001a, Malone et al. 2001b, Martin et al. 2002, Tran et al. 2002). However, in a review of ESRI’s library(ESRI 2002), containing its recent publication list from 1995 to 2002, health-related abstracts made up only 3.0% of the total, while environmental science comprised 11.4%. Clearly, efforts to integrate GIS into the health sciences must continue. Although health research is beginning to make use of GIS in mapping and analysis, we are unaware of any health-related studies that use GIS as a sampling tool. Public health surveys rely heavily on random sampling methods (Hennekens and Buring 1987). In the United States, this often means that researchers use random-digit dialling or stratified sampling based on street or block maps(Green and Krosnick 2001). However, achieving a random sample of interviewees is not as easy when working in the rural developing world. Often there are no reliable maps available, and telephone questionnaires are useless because the majority of the population has no telephone. Many rural villages are built haphazardly, without well-defined streets that investigators can use as markers for stratified sampling of homes. Therefore, the use of field maps developed using global positioning systems, which can later be downloaded into a GIS system such as ArcView, is immensely helpful. We report here on a method for developing random samples using a combination of ArcView mapping and field GPS. With this method, many of the sampling problems encountered by public health researchers in developing countries can be easily solved. Furthermore, the method is generalizible to other types of research, such as environmental sampling in difficult terrain.