RESUMO
To determine the value of an epidemiological surveillance system, we evaluated the incidence rate and trends of American Cutaneous Leishmaniasis (ACL) in three health districts of Ceara state, Brazil. We used a retrospective time series analysis of ACL notifications in Ceara state between 1986 and 1990. The trend of this endemic/epidemic disease was analysed by using three types of modeling on the moving average graph: a linear, linear periodic and non linear periodic model. The study samples consisted of the 14,615 cases of ACL notified in Ceara state between 1986 and 1990. Males accounted for 54% of cases. Age was between 1 month and 19 years in 42% of cases. Most cases (81%) were diagnosed on clinical grounds only. After calculation of the moving average and application of a linear regression model to these means, the trend analysis revealed that 1) there was a significant trend towards an increase in two of the three health districts, with correlation coefficients of 0.31 and 0.49 (Student's t test = 2.24 and 3.88) and a significant trend towards a decrease in the remaining district (correlation coefficient = 0.51, t = 4.1). The linear periodic model showed a cyclic representation with a 36-month period and a base of 241. The non linear periodic model showed a cycle of 32 months and a base of 241, and was a better representation of the disease variations. Despite the limited study period, it appears that the 32-month cycle could be attributed to meteorological conditions. Field studies are underway to determine the precise causes of this periodicity.