Health applications with GIS: Afyonkarahisar case

In this study, general information about the most common methods for clustering is given and the distribution of the Diabetes Mellitus Type 2 (DM), Hypertension (HT), Hyperlipidemia (HLD), Anemia (HB: low hemoglobin levels) and Thyroid diseases (TRD) in Afyonkarahisar province is analysed by using Geographical Information Systems and clustering focused statistical methods according to the results of health scanning.The data used in application is achieved with the health scanning between July 2005 and July 2006 at 18 townships icluding central township and 57 towns, totally 75 scanning regions. ArcGIS 9.0 (License no: 37157307 27004) and GeoDA 0.95-i (free) softwares are used for data analysis. The noise due to the different population areas with different disease case quantities and small numeric faults is eliminated with Emprical Bayes Method so prevalence ratio of diseases is smoothed. The places where these diseases in this research are mostly observed in Afyonkarahisar are determined, outlier values are detected with applying box plots, and information about the trends in the region is gained with spatial moving average based spatial rate analyses.Moran I and Getis-Ord General G methods are applied for determinig the spatial general clustering of the diseases in the research area, while Local Moran I ve Getis-Ord Gi* methods were used for local clustering analysis. Also Moran I based LISA which is one of the local clustering methods in GeoDA software is used. In the last stage of the application, prevelance ratios of the diseases are compared with the ratio of Turkey and similar conditioned provinces.According to the results of the analysis, the places are determined as townships where these diseases are most frequently observed, Bolvadin for DM, Başmakçı, Şuhut, Emirdağ and Sultandağ for HT, Başmakçı, Sandıklı and Emirdağ for HLD, Dazkırı, İscehisar, Çay and Emirdağ for HB, Dinar and Şuhut for TRD.Outlier places are detected with box plots analysis. In DM cases, Bolvadin for observed at males and observed at both males and females, in HLD cases, Merkez for observed at females, in HB cases, Dazkırı and Sinanpaşa for observed at males, Sinanpaşa and Çobanlar for observed at 19?40 age group, Çay, İscehisar and Dazkırı for observed at 41?64 age group, in TRD cases, Şuhut and Dinar for observed at females, observed at both males and females, observed at 19?40 age group, observed at 41?64 age group and observed at over 65 age group are determined as outlier townships. There is no detected outlier township for HT cases.According to the global clustering analysises, only HLD observed in men case is determined as statistacally meaningful with low values clustered in the region. Local clustering with low values at Merkez, İhsaniye, İscehisar ve Çobanlar for HLD cases at 41?64 age group commented as statistically meaningful. When the prevalence rates of these diseases at the region are compared with general Turkey and regions with similar geographic conditions, for HLD cases the rate is approximately equal to the Turkey average rate, for DM cases the rate is over the Turkey average rate, and for HT cases observed at over 41 age group the rate is very close to the Turkey average rate.If any meaningful clustering detected, the health expense will decrease and time will be gained in terms of propagation of the disease because of the decrement of target region dimensions with epidemiological researches. For this reason GIS and spatial statistic based applications must be used with an integrity in health sector. With this scope, completed study comes into prominence in terms of using the health policy efficeintly in the region for these diseases.Keywords: Spatial Autocorrelation, GIS, Spatial Statistics, Diabetes Mellitus, Hypertension, Hyperlipidemia, Anemia, Thyroid disease