Evaluation of spatial cluster detection methods for dengue fever in the state of Paraiba, Brazil.

Abstract

This study is a quantitative, ecological, descriptive, retrospective, cross-sectional study on dengue in the state of Paraíba in north-eastern Brazil aimed to compare the performance of spatial clustering methods based on epidemiological data. The population consisted of all people residing in the state, and the sample was all dengue fever cases reported annually between 2018 and 2022. The residence localization of people suffering from dengue fever was used to identify the spatial distribution of this infection in the Paraíba State. Scan Statistics, Besag-Newell, Getis-Ord, MStatistics and Tango were used and it was observed that the methods Getis-Ord, M-Statistic and Tango showed large spatial clusters, which included municipalities with high and low values. Scan Statistics and Besag-Newell's method also showed most of these clusters, with Scan Statistic providing better agreement with the high Standardized Incidence Ratio (SIR) than Besag-Newell's method. In conclusion, Scan statistic outperformed the other methods by identifying significant clusters in greater proportion in all study periods when mapping using Rigorous Impact Evaluation (RIE) was applied. However, it is necessary to consider each method's assumptions to select the most appropriate method for each application. Thus, this study provides relevant elements to help decision makers manage and prevent diseases, such as dengue fever and other vector-borne diseases.