Abstract

SPATIAL ANALYSIS OF CANCER INCIDENCE IN AFRICA

Omolehin J.O1, Korter G.O2, Abolarin O.E3

---

Abstract Spatial data analysis has brought new power to the analysis of spatial data, capturing effects of redundant locational information contained in cross sectional data. The objective was to investigate the spatial pattern of cancer incidence across geographical regions in Africa. The focus was on African data in 61 countries for 2012 estimated incidence of all cancers (excluding melanoma skin cancer incidence) for all ages and both sexes. The Moran?s and Getis and Ord?s statistics were used for spatial analysis. Moran?s and Getis-Ord indices equal 0.02 and 0.03 while Z score equal 0.84 and 0.03 standard deviations respectively. There was no apparent clustering detected at this scale. The null hypothesis of randomness was not rejected. This means the expected pattern was just one of the many (n!) possible versions of complete spatial randomness of the distribution of cancer incidence. Therefore, it was concluded there was no spatial dependence of cancer incidence across Africa. Ultimately, government policies on preventive measures against cancer will be country specific. The proposed statistical technique is more informative with the advantage of providing geo visual representations of results, thus giving a better understanding of parameter estimates. Key words: Cancer, Spatial Statistics, Locational Information, Incidence

PDF