Spatio-temporal clustering analysis using two different scanning windows: A case study of dengue fever in Peninsular Malaysia.

Dengue fever is a mosquito-borne viral infection of humans caused by a virus of the Flaviviridae family. In Malaysia the annual incidence risk of dengue fever for the period 2000 to 2019 ranged from 30 to 390 cases per 100,000. The aim of this paper was to identify spatial, temporal and spatio-temporal clusters of dengue fever in Peninsular Malaysia for the period 2015 to 2017. Counts of confirmed incident cases of dengue fever for each of the 86 districts of Peninsular Malaysia for the period 1 January 2015 to 31 December 2017 (inclusive) and district-level census data allowed us to calculate the incidence rate of dengue fever, defined as the number of confirmed cases of dengue fever per 100,000 person-years at risk. We applied Kulldorff's cylindrical space-time scan statistic and Takahashi et al.'s prismatic space-time scan statistic to the data. We identified no major differences in the number and location of spatial clusters of dengue incidence for 2015, 2016 and 2017 using Kulldorff's and Takahashi et al.'s method. Spatio-temporal clusters of dengue occurred at several times throughout each year in various high population dense areas. These clusters not only included high population density districts but also their adjacent district neighbours. The temporal clustering of dengue cases during the monsoon season (mid September to late December each year) implies that there is a biologically plausible causal association between rainfall and the incidence of dengue. Identification of locations and time periods when the frequency of dengue is high allows Malaysian public health authorities to be more objective in their decision making around vector control and dengue public awareness campaigns. Future research will quantify the association between population density and rainfall on dengue incidence. This will allow health authorities to take a more proactive approach for dengue control.