The Urban Spread of Visceral Leishmaniasis: Clues from Spatial Analysis Guilherme L. Werneck,1'2 CarlosH. N. Costa,3 Alexander M. Walker,4 John R. David,1 Matthew Wand,5 and James H. Maguire1 Background. The pattern of spread of visceral leishmaniasis in Brazilian cities is poorly understood. Methods. We used geographic information systems and spatial statistics to evaluate the distribution of 1061cases of visceral leishmaniasis in Teresina, Brazil, in 1993 through 1996. Results. A locally weighted (LOESS) regression model, which wasfit as a smoothed function of spatial coordinates, demon- strated large-scale variation, with high incidence rates in pe- ripheral neighborhoods thatbordered forest land andpastures. Moran's I indicated small-scale variation and clustering up to 300 m, roughly the flightrange of the sand fly vector. Conclusions. Spatial analytical techniques can identify high- risk areas fortargeting control interventions. (EPIDEMIOLOGY 2002;13:364-367) Key words: visceral leishmaniasis, spatial analysis, smoothing, clustering, geographic information systems, infectious disease, Brazil. orldwide,400,000 cases of visceral leishman- iasis (VL) with 50,000 fatalities occur annu- ally.1 Over 90% of cases in the Americasarise in Brazil,where the protozoan Leishmania chagasi is the responsibleagent.2'3 Recently, several Brazilian cities of over 500,000 inhabitants experienced epidemics of this typically ruraldisease as a result of massive population movements and rapid,uncontrolled urbanization.4'6'7 Ef- forts to control the canine reservoir host and the phle- botomine vector, Lutzomyia longipalpis, failed to inter- rupt transmission and prevent new epidemics.4'5'8 The broad geographic dimensions of Brazilian cities make control interventions logistically and economically dif- ficult to sustain. Although foci of high incidence are readilyidentified in ruralareas,9 little is known about the distributionof cases in densely populated urban areas, where social Fromthe IDepartment of Immunology and InfectiousDisease, Harvard School of Public Health, Boston, MA; 2Department of Preventive Medicine/NESC, Uni- versidade Federal do Rio de Janeiro,Rio de Janeiro,Brazil; 3Hospital de Doen9as Infecto-Contagiosas, Universidade Federaldo Piaui, Teresina, Piauf, Brazil; and 4Department of Epidemiology and 5Department of Biostatistics,Harvard School of Public Health, Boston, MA. Address correspondenceto: James H. Maguire,ParasiticDiseases Epidemiology Branch, Centers for Disease Control and Prevention, 4770 Buford Highway, N.E., Atlanta, GA 30341-3724; jmaguire@cdc.gov This workwas supported in partby a grantfromthe National Institutesof Health (AI-16303-11) and the Coordenacao de Aperfeicoamento de Pessoal de Nivel Superior (CAPES/MEC, Brazil). Submitted 23 April 2001; final version accepted 18 January 2002. Copyright ? 2002 by Lippincott Williams & Wilkins, Inc. networks and interactions between human settlements and the naturalenvironment are more varied and com- plex. We therefore sought to determine spatial patterns of disease in an urbansite and to identify high-riskareas for efficient targeting of control measures. We used geographic information systems (GIS) and spatial statistical techniques to identify patterns of the occurrence of VL during an urban epidemic. To accu- rately define these patterns it was necessaryto examine both large-scale and small-scale variation in incidence rates.10?11 Large-scale variation, or spatial gradient,refers to the trend in the mean value of a process (eg, inci- dence) in space and can be represented as a function of geographical coordinates.12 Small-scalevariation,or sec- ond order effect, results from clustering of high or low values across the region and from spatialautocorrelation, the property by which nearbyareastend to have similar values of a variable.11,13 To measure small-scale varia- tion, it is helpful first to remove the effects of large-scale variation.1' Methods Study Area Teresina, the capital of Piauf State, lies at 05005' South latitude at the convergence of the Pamaiba and Poti rivers (Figure1). Shrubs and sparse mango and palm trees are the predominantvegetation in the city, which is surrounded by tropical forest and pastureland. Teresinawas the site of Brazil's firsturban epidemic of VL4 in 1980-1985, when approximately 1000 cases oc- 364