Vulnerability to the transmission of human visceral leishmaniasis in a Brazilian urban area

ABSTRACT OBJECTIVE To analyze the determinants for the occurrence of human visceral leishmaniasis linked to the conditions of vulnerability. METHODS This is an ecological study, whose spatial analysis unit was the Territorial Analysis Unit in Araguaína, State of Tocantins, Brazil, from 2007 to 2012. We have carried out an analysis of the sociodemographic and urban infrastructure situation of the municipality. Normalized primary indicators were calculated and used to construct the indicators of vulnerability of the social structure, household structure, and urban infrastructure. From them, we have composed a vulnerability index. Kernel density estimation was used to evaluate the density of cases of human visceral leishmaniasis, based on the coordinates of the cases. Bivariate global Moran’s I was used to verify the existence of spatial autocorrelation between the incidence of human visceral leishmaniasis and the indicators and index of vulnerability. Bivariate local Moran’s I was used to identify spatial clusters. RESULTS We have observed a pattern of centrifugal spread of human visceral leishmaniasis in the municipality, where outbreaks of the disease have progressively reached central and peri-urban areas. There has been no correlation between higher incidences of human visceral leishmaniasis and worse living conditions. Statistically significant clusters have been observed between the incidences of human visceral leishmaniasis in both periods analyzed (2007 to 2009 and 2010 to 2012) and the indicators and index of vulnerability. CONCLUSIONS The environment in circumscribed areas helps as protection factor or increases the local vulnerability to the occurrence of human visceral leishmaniasis. The use of methodology that analyzes the conditions of life of the population and the spatial distribution of human visceral leishmaniasis is essential to identify the most vulnerable areas to the spread/maintenance of the disease.


INTRODUCTION
Visceral leishmaniasis (VL) is a vector-borne parasitic disease whose importance in the context of public health in Brazil has increased significantly in recent years. This fact is mainly due to the expansion of urbanization processes and the modification of habitats of the species involved in the transmission cycle 5,22 .
The geographical areas with transmission of VL are increasingly expanding. The disease, considered one of the priorities among tropical diseases by the World Health Organization (WHO), is endemic in ninety-eight countries, including Brazil a , whose areas of occurrence increase progressively 9,13 . There is a great gap in scientific knowledge about VL. Studies are needed on new drugs and therapeutic regimes, as well as studies that address the effectiveness of disease control actions with greater methodological strength, among others aspects 22 .
Araguaína, in the State of Tocantins, Brazil, presents a recent endemic-epidemic process of VL. There has been an alarming increase in the number of confirmed human cases mainly between 2006 (56) and 2007 (251), when it presented the greatest number of records of the disease in Brazil (7.0% of all reported cases in the country). Since then, it remains among the four Brazilian municipalities with the highest absolute number of annual records of VL.
The analysis of the vulnerability 20 of the region for the transmission of VL using indicators and the analysis of the spatial distribution of urban VL can contribute to the better understanding of the heterogeneous transmission process, resulting from the socio-territorial organization of the municipality 1,2,5 .
This study aimed to analyze the determining circumstances for the occurrence of human visceral leishmaniasis (HVL), linked to the conditions of vulnerability in Araguaína. The data can provide elements for the formulation of public policies adapted to this situation. Figure 1) is located in northern Brazil, in the northern portion of the State of Tocantins. It has an altitude of 227 m, under the geographical coordinates 7°11'28" South latitude and 48°12'26" West longitude. It is 380 km from Palmas, capital of the State (IBGE, 2014). It presents high temperatures throughout the year, approximately 28°C. The population distribution is predominantly urban, with great rural emptiness. Between the 2000 and 2010 Demographic Censuses (Brazilian Institute of Geography and Statistics -IBGE) b,c , we can observe a growth of 33% ( from 113,143 to 150,484 inhabitants) of the total population of the municipality and approximately 35% of the urban population ( from 105,874 to 142,925 inhabitants).

Araguaína (
The urban territory of Araguaína is characterized by a central region with high density and intermediate and peripheral areas, discontinuous among themselves. The growth model of this municipality is discontinuous and segregated. For the most part, the areas with increased population density have vast unoccupied areas with considerable spacing between the buildings. In addition, most of the streets in these areas are not paved. We performed the analysis of the situation of vulnerability to the transmission of the disease based on the construction of indicators related to the conditions of urban, social, and household infrastructure for 2010, the analysis of the distribution of HVL in time and space in two three-year periods (2007-2009 and 2010-2012), in addition to the descriptive analysis of the cases of canine visceral leishmaniasis (CVL) notified by the Municipal Health Department between 2006 and 2008. The analysis in three-year periods was used because the disease has a cyclical nature. In addition, this approach helps to reduce the variability in the number of cases reported between the years of the study.
The digital base of the territorial analysis units (TAU) was made from the compatibility of the bases of urban census tracts of the 2000 and 2010 Demographic Censuses of the IBGE d . The compatibility was carried out because those geographical bases are very distinct ( Figure 1).
We generated polygons of the areas occupied by the human population, based on the methodology referred to in the Land Use Manual 11 . The polygons of the occupied areas were vectorized by the method of visual interpretation of satellite images. We identified the spaces with urban characteristics at a 1:5,000 scale, using geoprocessing tools in a Georeferenced Information System (GIS). This step was developed based on the Digital Globe satellite images for 2014, obtained by the software Google Earth, with spatial resolution of one meter. The software used was ArcGis 10.2.
The autochthonous cases of HVL reported between 2007 and 2012 were georeferenced from the address found in the database of the SINAN, with the use of Google Earth. This generated a digital data mesh.
The mesh containing the specific data relating to the cases of the disease was superimposed to the digital base of TAU and the areas occupied by human populations in each TAU. This allowed the aggregation of cases of HVL based on these units of territorial analysis.  Kernel density estimation was used to evaluate the density of cases of HVL, based on the coordinates of the cases. Thematic maps were prepared for the spatial and temporal analysis of the evolution of the endemic disease, as well as their correlation with the information about the situation of social and household structure and urban infrastructure. The application ArcGis 10.2 was used for this analysis.
Official information, available on the website of IBGE on the 2010 Demographic Census, was used for the analysis of the conditions of urban, social, and household infrastructure of the municipality. This analysis identified the most vulnerable areas in the urban and social infrastructure, in order to reveal differences between the various areas of the municipality.  Table 1). The composite vulnerability index (CVI) was prepared based on these indicators  The IVSS was created to combine the information from various sociodemographic indicators into a synthetic indicator to establish an order in relation to the level of social vulnerability. The IVHS was prepared with the information of various indicators of household structure, allowing a synthetic measure of the housing conditions of the population. The IVUI was created to synthesize the information on the urban infrastructure surrounding the household into a single indicator.
Bivariate Global Moran's I was used to verify the existence of spatial autocorrelation between the incidence rate of HVL and the Vulnerability Indices. The index tests whether the values of an indicator in a given region is related to the values of another variable in neighboring regions. Statistics can have the values from -1 to +1, positive for direct correlation and negative when inverse 3,16,e . Bivariate Global Moran's I does not show where the spatial cluster are. Thus, we used Bivariate Local Moran's I, which provides the degree of spatial autocorrelation, statistically significant, in each regional unit. According to Anselin, this index provides an indication of the degree of linear association between the value for a variable in a given area and the average of another variable in the surrounding areas. The matrix of the neighborhood W defined by contiguity was the Queen matrix, which considers two neighboring regions that have common borders, in addition to common nodes (vertices).
The statistical significance of Moran's I Index was tested using permutations. We generated subsequent permutations of the values of the data associated with the TAU. Each permutation produces a new spatial arrangement, in which values are distributed randomly between the TAU. That way, we can create an empirical distribution of statistics I under the null hypothesis of spatial randomness. The empirical pseudo-significance was based on ninety-nine random permutations. The program used in this analysis was GeoDa 1.6.6 3 .
The information about the cases of CVL was provided according to neighborhood by the Center of Control of Zoonosis (CCZ) of Araguaína. As there was incompatibility of the sketch of neighborhoods with the digital cartographic bases of the census tracts and TAU, the analyses were restricted to the municipal level. We calculated the positivity of cases of CVL from the absolute number of samples collected and the number of samples per year.
This study has been approved by the Research Ethics Committee of the Escola Nacional de Saúde Pública Sérgio Arouca of the Fundação Oswaldo Cruz (Protocol 142/11).

RESULTS
We observed an increase of approximately 30% of the resident population in the urban area of the municipality of Araguaína between the 2000 and 2010 Censuses. We could also identify a more marked growth (6.7%) in the population aged 60 years or more and a less accentuated growth (5.9%) among individuals aged between zero and four years.   Table 2).  Bivariate Local Moran's I showed the presence of statistically significant clusters between the incidences of HVL in both periods and the indicators of vulnerability for the census tracts ( Figure 4).

Main clusters (Low-Low in central areas and High-High in the northern and southern peripheral regions) were evidenced, as well as intermediate clusters (Low-High and
High-Low) in all combinations tested in the period analyzed ( Figure 4). The main clusters located in the central region showed areas of low incidence of HVL circumscribed by better living conditions (those with low scores), while the peripheral clusters revealed areas of high incidence of HVL circumscribed by areas with worse living conditions (those with high scores). The High-High clusters located in the southern peripheral region of the municipality remained virtually unchanged in both periods for the indicators IVUI and CVI (Figure 4).

DISCUSSION
We observed a pattern of centrifugal spread of human visceral leishmaniasis in Araguaína.
Outbreaks of the disease reached progressively, from 2008, the occupied central and peri-urban areas of the municipality. However, the pattern of spread had important annual intensity variations.
Other Brazilian municipalities, also in the process of increasing urbanization and precarious living conditions, have different epidemiological patterns of spread of VL. Among these patterns, we can identify: restricted peri-urban pattern (outbreaks of the disease restricted to areas of the urban periphery), diffuse peri-urban pattern (outbreaks of the disease mainly affecting the urban peripheries and, secondarily, central/urbanized areas), epidemiological pattern of centrifugal spread (outbreaks of the disease spread within the municipality, and also, increasingly, in neighboring municipalities), and, finally, pattern of network dissemination (outbreaks of the disease sequentially affecting interconnected municipalities) 1,4,6,13,18 . Few municipalities have a rural epidemiological pattern, dominated by rural outbreaks with occurrence of sporadic human cases 19 .
The centrifugal process of spread of the disease in Araguaína is characterized by important changes in the territory, such as: disorderly expansion of the city -which has advanced through the natural habitat of the vector of VL -, lack of basic and sanitary infrastructure -which enable the urbanization of the vector -, and the spread of the disease in the municipality 5,7,17,18,21 .
The increase of approximately 30% of the population living in the municipality between the 2000 and 2010 Censuses and the changes of the urban space that are inherent to it can help to explain the maintenance of the high number of human cases of VL notified in the historical series (2007-2012), although we can observe a wide variation in the number of reported cases over the years. There was an increase of susceptible and vulnerable persons inserted in that territory, as well as individuals -migrants and their animals, for example -from endemic areas for VL 8,14,f .
Although the notification of cases of VL is the main strategy for surveillance of the disease in Brazil, the process of notification and confirmation of cases can be slow. The HVL is considered a mandatory notifiable disease in the country since 1975. Its vigilance is decentralized and especially uses the data of notifications recorded in the SINAN. The analysis of these data allows the space-temporal monitoring of the disease in the country, subsidizing the actions for its control 15 .
A correlation was observed between higher incidences of HVL and worse living conditions, although it was not high. Although the TAU presented certain uniformity regarding the indicators analyzed, we could observe clusters from the calculation of Bivariate Local Moran' s I. This approach identified that the environment in circumscribed areas contributes as a factor of protection or increase for the relative local vulnerability concerning HVL in specific locations in Araguaína 17 . This reinforces the theory that the occurrence of HVL is closely related to the situation of social vulnerability g in which parts of the population are inserted 18,g .
The use of indicators of vulnerability regarding the territory contributed to the knowledge on the intra-urban differences more associated with the occurrence of HVL. The analysis of spatial variability based on the combination of these indicators of vulnerability and the incidence of VL allowed the identification of particularly distinct areas, i.e. with their own socioeconomic characteristics that can contribute to the process of reproduction of the disease in the municipality. Such areas should be the target of local public policies aiming to reduce the vulnerability and decrease the number of cases of the disease in the territory of Araguaína 20,h .
The occupation of the urban space arises from the actions of society on the nature. Its configuration incorporates social and economic structure and its dynamics, establishing the flows of the local movement of goods and services. The lack of consistent data about the canine reservoir is one of the limitations of the study. We highlight the annual variation of the areas included in the canine serological surveys conducted by the CCZ of Araguaína and the lack of regularity of the survey in each area for the years of study. The variation of the number of dogs included in the surveys annually has not allowed further evaluations on the seroprevalence of VL.
The scenario of visceral leishmaniasis in Brazil is complex. Researchers need to seek new methodologies of analysis that can cover the various elements involved in the process of introduction, dissemination, and maintenance of the disease. The difficulties related to its control must be considered, from the control of the canine population and vectors and the identification of infected animals up to the elimination of risk factors.