CLIMATIC AND DEMOGRAPHIC DETERMINANTS OF AMERICAN VISCERAL LEISHMANIASIS IN NORTHEASTERN BRAZIL USING REMOTE SENSING TECHNOLOGY FOR ENVIRONMENTAL CATEGORIZATION OF RAIN AND REGION INFLUENCES ON LEISHMANIASIS R. ALEX THOMPSON, JOSÉ WELLINGTON DE OLIVEIRA LIMA, JAMES H. MAGUIRE, DEWITT H. BRAUD, AND DANIEL T. SCHOLL Department of Pathobiological Sciences, School of Veterinary Medicine, and Department of Geography and Anthropology, Louisiana State University, Baton Rouge, Louisiana; Fundação National de Saúde/Ceará, Fortaleza, Ceará, Brazil; Department of Immunology and Infectious Disease, Harvard School of Public Health, Boston, Massachusetts Abstract. Remote sensing (RS) permits evaluation of spatial and temporal variables that can be used for vector- borne disease models. A Landsat Thematic Mapper scene covering Canindé, Ceará in northeastern Brazil (September 25, 1986) was spectrally enhanced and classified using ERDAS (Atlanta, GA) Imagine for 873 4-km 2 areas. The population and number of cases of American visceral leishmaniasis (AVL) were determined for each 4-km 2 area. Relative risk (RR) ratios were calculated for climate, demographic, and case data recorded for 17 years by the Munici- pality of Conide ´ The RR of AVL for a child less than 10 years old from the foothills relative to non-foothill residency was 4.0 (95% confidence limit  3.5, 4.5). The RR of AVL in children was 9.1 during a time when the three-year rolling rain average (current year plus two previous year’s precipitation) was between 40 and 60 cm relative to rain greater than 100 cm. The results suggest that features detected by RS techniques combined with climatic variables can be used to determine the risk of AVL in northeastern Brazil. INTRODUCTION Historically, 90% of the cases of American visceral leish- maniasis (AVL) in Brazil occurs in the northeastern part of this country, where the highest incidence occurs among chil- dren living in foothills of rural areas. 1,2 The protozoa respon- sible for this disease, Leishmania chagasi, which can be iso- lated from humans, dogs, and foxes, is transmitted by the bite of the female phlebotomine sand fly Lutzomyia longipalpis (Diptera: Psychodidae). 3-5 Symptoms of AVL (kala azar) de- velop after a 2-4-month incubation period, but 6-18 times more people experience asymptomatic infections than symp- tomatic infections. 6 The treatment of choice is a 28-day course of pentavalent antimony (Sb 5 ). Without treatment case fatal- ity approaches 95%. 7 The sand fly vector is nocturnal, anthropophilic, reproduces best between 23°C and 28°C and a relative humidity of 70-100%, and has a short flight range. 8 It completes its life cycle in 5-8 weeks under ideal conditions. 9 Such specific en- vironmental conditions required for vector propagation make leishmaniasis an ideal candidate for study using remote sens- ing (RS) by satellite. 10,11 Remote sensing, image processing, and geographic infor- mation systems (GIS) are computer programs designed to collect, store, and analyze data relative to geographic loca- tions. They are being used more frequently in public health studies to identify, classify, and organize environmental vari- ables that influence vector distribution and abundance. 12 Sat- ellite-borne sensors measure reflected electromagnetic and emitted thermal energy, and algorithms are available to clas- sify data into surrogate indices for surface vegetation, mois- ture, or dryness. 11,13 These indices are used in GIS and in statistical models to define associations between climate and the incidence of vector-borne disease. Investigations have used RS/GIS technology to study leishmaniasis and climate in the Middle East, the ecology of malaria in sub-Saharan Af- rica, environmental determinants of Rift Valley Fever virus infection in Kenya, and risk for Lyme disease in the United States. 11,13-19 In this study, RS and GIS were used to separate the envi- ronment into foothills or plains categories with electromag- netic spectral analysis of vegetation to evaluate the effect of age and precipitation by region on the incidence of AVL in the municipality of Canindé (Ceará State), Brazil. MATERIALS AND METHODS Study area. The municipality of Canindé, with an area of 2,883 km 2 , is located 113 km inland from Fortaleza, the coastal capital of Ceará, Brazil. Canindé, the administrative center of the municipality, is situated at 4°27 ' 32  S, 39°18'32W. It has an altitude of 150 meters above sea level and is situated on a plain between small mountains in the municipality of Baturité and the foothills of Canindé. The highest peak in Canindé is 1,085 meters above sea level, and most of the plateaus in the foothills range between 300 and 600 meters above sea level. The climate is biseasonal with a seven-month dry season and a five-month rainy season during normal years. The mean daily maximum and minimum temperatures are 30°C and 24°C, respectively, with larger diurnal variations than the variations in mean monthly temperatures. The annual aver- age rainfall is 756 mm, ranging from a minimum of 191 mm in 1993 to the maximum of 1,678 mm in 1985. 20,21 The popula- tion of the municipality increased from 50,652 people in 1970 to 61,650 people in 1991, during which time there was exten- sive migration from the rural zone to the cities Canindé and Fortaleza. 20 There has been little immigration into the mu- nicipality from other regions, and it is estimated that 98% of the residents of Ceará live in their municipality of birth. 20 Study design. A retrospective cohort study encompassing 17 years was designed in which the units of observation were 2 km × 2 km (4 km 2 )-inhabited areas of land within the mu- nicipality. This unit of observation was chosen because of the short flight range of sand flies, the spatial resolution (30 m 2 ) of Landsat Thematic Mapper (TM) satellite scenes, the 2 km × 2 km resolution of the 1:100,000 universal transverse mer- Am. J. Trop. Med. Hyg., 67(6), 2002, pp. 648–655 Copyright © 2002 by The American Society of Tropical Medicine and Hygiene 648