The objectives of this study were to study the epidemic patterns of dengue hemorrhagic fever and other acute febrile illnesses in Yala province in southern Thailand, to investigate relations between the their incidence rates in terms of their geographical distributions, and to develop a methodology that may be applied routinely to geographical epidemiologic research for the spatio-temporal mapping of disease. Schematic range maps and statistical models were used to investigate their distribution by year and location. The analytic methods used poisson and negative binomial distribution models. The concept of a "risk alert" is suggested as a method for highlighting subdistricts with unexpectedly high incidence rates in a given year.
We found high correlations between the incidence rates in 2002 for diarrhoea and conjunctivitis (r = 0.805), and between diarrhoea and pneumonia (r = 0.798). For all illnesses in the two years, five risk alerts occurred. For dengue fever, in 2002 there were two subdistricts with moderate risk alerts, namely KaYuBoKo (p = 0.030) and KuTaBaRu (p = 0.042), whereas in 2003 there was just one subdistrict with a high risk alert, KaYuBoKo (p = 0.003). For pyrexia, in 2003 there were two subdistricts with moderate risk alerts, namely Kerikat (p = 0.017) and ThanTo (p = 0.036).
Further studies are needed, over longer periods, involving more provinces, and taking into account known risk factors including age, season, environmental factors, and time series autocorrelations.