Abstract
Tuberculosis is one of the most contagious diseases. Although it has been present for over 5000 years, it is still one of the most significant public health problems. Tuberculosis is defined as a disease caused by a bacterium called Mycobacterium tuberculosis. Its incidence has increased and decreased over time, but it has always been a constant threat to public health. In the 1940’s some medicines were discovered to treat tuberculosis, so it slowly began to decrease. But then countries let their guard down, tuberculosis was neglected and so cases increased.

Understanding spatial relationships between determinants and outcomes of health care is important as the concept of population-based health care gains acceptance. A wide range of tools for understanding these spatial relationships is available to the family medicine researcher through the use of Geographic Information Systems (GIS). The power of GIS lies in its ability to display the spatial distribution of a health related predictor or outcome. These maps can then be used to either generate or test hypotheses that would not have otherwise occurred to the investigator without visualizing the spatial relationships. The type of GIS application used is dependent on the type of data the researcher has and the research question. The three most common types of data are point or event data, lattice data, and geostatistical data.

Abstract
Background: Various geographic information systems (GIS) are now widely used to map the distribution of diseases and mortality. However, the mapping of raw mortality rates has been found to be inappropriate since it does not account for the spatial heterogeneity of the population at risk. Bayesian techniques have therefore been suggested as a solution to the problem. Methods: Annual mortality rates for each of the 39 villages of the study area in the Kossi Province in northwest Burkina Faso were calculated using midyear populations of children under five. Two mapping techniques were then used. Firstly, the GIS software ArcView was used to map the crude mortality rates. Secondly, the data were smoothed by the method of empirical Bayes estimation. The geostatistical prediction method of Kriging was then used to spatially interpolate the data for successive years. Results: No spatial pattern is identifiable from the circles representing mortality rates drawn on the map using ArcView. The circles are scattered over the study area and comparing annual distributions between them is difficult. The maps produced by the Bayesian technique also do not show a clear spatial trend pattern. However, they indicate the tendency of villages in the northeastern region to produce higher incidence or risk values, confirming the results of an earlier study reporting a significant cluster of high childhood mortality in the same area.