SEARCH

SEARCH BY CITATION

Keywords:

  • leprosy;
  • control;
  • public health;
  • spatial analysis;
  • Brazil;
  • epidemiology
  • lèpre;
  • contrôle;
  • santé publique;
  • analyse spatiale;
  • Brésil;
  • épidémiologie
  • Lepra;
  • control;
  • salud pública;
  • análisis espacial;
  • Brasil;
  • epidemiología

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Objective  The Brazilian National Hansen’s Disease Control Program recently identified clusters with high disease transmission. Herein, we present different spatial analytical approaches to define highly vulnerable areas in one of these clusters.

Method  The study area included 373 municipalities in the four Brazilian states Maranhão, Pará, Tocantins and Piauí. Spatial analysis was based on municipalities as the observation unit, considering the following disease indicators: (i) rate of new cases/100 000 population, (ii) rate of cases <15 years/100 000 population, (iii) new cases with grade-2 disability/100 000 population and (iv) proportion of new cases with grade-2 disabilities. We performed descriptive spatial analysis, local empirical Bayesian analysis and spatial scan statistic.

Results  A total of 254 (68.0%) municipalities were classified as hyperendemic (mean annual detection rates >40 cases/100 000 inhabitants). There was a concentration of municipalities with higher detection rates in Pará and in the center of Maranhão. Spatial scan statistic identified 23 likely clusters of new leprosy case detection rates, most of them localized in these two states. These clusters included only 32% of the total population, but 55.4% of new leprosy cases. We also identified 16 significant clusters for the detection rate <15 years and 11 likely clusters of new cases with grade-2. Several clusters of new cases with grade-2/population overlap with those of new cases detection and detection of children <15 years of age. The proportion of new cases with grade-2 did not reveal any significant clusters.

Conclusions  Several municipality clusters for high leprosy transmission and late diagnosis were identified in an endemic area using different statistical approaches. Spatial scan statistic is adequate to validate and confirm high-risk leprosy areas for transmission and late diagnosis, identified using descriptive spatial analysis and using local empirical Bayesian method. National and State leprosy control programs urgently need to intensify control actions in these highly vulnerable municipalities.

Objectif:  Le Programme national brésilien de lutte contre la maladie de Hansen a récemment identifié des groupes avec une transmission élevée de la maladie. Nous présentons ici différentes approches analytiques spatiales pour définir les zones hautement vulnérables dans un de ces groupes.

Méthode:  La zone d’étude comprenait 373 municipalités dans quatre états brésiliens: Maranhão, Pará, Tocantins et Piauí. L’analyse spatiale a été basée sur les municipalités comme unités d’observation, compte tenu des indicateurs suivants de la maladie: (1) Taux de nouveaux cas/100.000 habitants, (2) taux de cas <15 ans/100.000 habitants, (3) nouveaux cas avec l’handicap de grade-2 /100.000 habitants et (4) proportion de nouveaux cas avec le handicap de grade-2. Nous avons effectué une analyse spatiale descriptive, une analyse bayésienne empirique locale et des statistiques sur l’analyse spatiale.

Résultats:  254 (68,0%) municipalités ont été classées comme hyper endémiques (taux de détection moyenne annuelle >40 cas/100.000 habitants). Il y avait un regroupement de municipalités avec des taux de détection plus élevés à Pará et dans le centre de Maranhão. Les statistiques sur l’analyse spatiale ont identifié 23 grappes probables de nouveaux taux de détection de cas de lèpre, la plupart d’entre eux localisés dans ces deux états. Ces grappes ne comprenaient que 32% de la population totale, mais 55,4% des nouveaux cas de lèpre. Nous avons également identifié 16 grappes importantes pour le taux de détection de cas <15 ans et 11 grappes probables de nouveaux cas avec l’handicap de grade-2. Plusieurs grappes de nouveaux cas avec le grade-2/population se chevauchaient avec celles de la détection de nouveaux cas et la détection des enfants <15 ans. La proportion de nouveaux cas avec l’handicap de grade-2 n’a pas révélé d’importantes grappes.

Conclusions:  Plusieurs regroupements de municipalités avec une transmission élevée de la lèpre et un diagnostic tardif ont été identifiés dans une zone d’endémie, par différentes approches statistiques. Les statistiques sur l’analyse spatiale sont adéquates pour valider et confirmer les zones de lèpre à haut risque pour la transmission et le diagnostic tardif, identifiées par l’analyse spatiale descriptive et par la méthode bayésienne empirique locale. Les programmes nationaux et étatiques de lutte contre la lèpre devraient urgemment intensifier les actions de contrôle dans ces municipalités très vulnérables.

Objetivo:  El programa Nacional de Control de la enfermedad de Hansen, en el Brasil, identificó recientemente conglomerados con niveles altos de transmisión. Presentamos diferentes enfoques del análisis espacial para definir áreas con alta vulnerabilidad en uno de estos conglomerados.

Método:  El área de estudio incluía 373 municipalidades en los cuatro estados Brasileros de Maranhão, Pará, Tocantins y Piauí. El análisis espacial estaba basado en municipalidades como unidad de observación, considerando los siguientes indicadores de enfermedad: (1) tasa de nuevos casos /100,000 habitantes; (2) tasa de casos <15 años/100,000 habitantes; (3) nuevos casos con discapacidad de grado-2 /100,000 habitantes; y (4) proporción de nuevos casos con discapacidad de grado-2. Hemos realizado un análisis espacial descriptivo, análisis Bayesiano empírico local y estadística de barrido espacial.

Resultados:  254 (68.0%) municipios fueron clasificados como hiperendémicos (tasa media de detección anual >40 casos/100,000 habitantes). La concentración de municipios con mayores tasas de detección estaban en Pará y el centro de Maranhão. La estadística de barrido espacial identificó 23 posibles conglomerados de tasas de detección de nuevos casos, la mayoría localizados en estos dos estados. Estos conglomerados incluían solo un 32% de la población total, pero 55.4% de los nuevos casos de lepra. También hemos identificado 16 conglomerados significativos para la tasa de detección <15 años y 11 posibles conglomerados de nuevos casos con grado-2. Varios conglomerados de nuevos casos de grado-2/habitantes se sobreponían con aquellos con detección de nuevos casos y detección de niños <15 años de edad. La proporción de nuevos casos con grado-2 no reveló conglomerados significativos.

Conclusiones:  Mediante el uso de diferentes enfoques estadísticos se identificaron varios conglomerados municipales con alta transmisión de lepra y diagnóstico tardío en un área endémica. La estadística de barrido espacial es adecuada para validar y confirmar áreas con alto riesgo transmisión de lepra y diagnóstico tardío, identificados mediante análisis espacial descriptivo y el método Bayesiano empírico local. Los programas nacionales y estatales de control de lepra necesitan urgentemente intensificar las acciones de control en estos municipios altamente vulnerables.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Leprosy remains a public health problem in many countries (Rodrigues & Lockwood 2011). Considering the limited availability of financial resources for leprosy research, the few financial rewards for the private sector and the fact that the poor are at higher risk for infection and clinical disease (Kerr-Pontes et al. 2006; Feenstra et al. 2011), leprosy is a Neglected Tropical Disease (Holveck et al. 2007).

After India, Brazil has the second-largest number of new cases of leprosy. In 2009, about 38 000 new cases were detected in the country, constituting almost 98% of all cases in the Americas (World Health Organization 2010). To intensify disease control measures on certain geographic areas, the Brazilian National Hansen’s Disease Control Program (Programa Nacional de Controle da Hanseníase – PNCH) recently identified the 10 most likely geographic clusters of high disease transmission and increasingly focussed control activities on these areas (Ministry of Health of Brazil 2008; Penna et al. 2009a). These clusters comprise about 51% of Brazil’s newly detected cases, but only 15.4% of the population (Ministry of Health of Brazil 2008).

Herein, we performed three methods of spatial analysis to define and classify high-risk areas in a major leprosy cluster. Priority regions for control of transmission and late diagnosis were identified.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We included all municipalities in the leprosy disease cluster in Brazil with greatest geographic extension, defined by the National Hansen’s Disease Control Program (Ministry of Health of Brazil 2008) (Figure 1). This cluster has an estimated population (2009) of about 10 million (5.2% of Brazil’s population), but 20% of the nation’s new leprosy cases. The demographic characteristics of the cluster have been described in detail previously (Alencar et al. 2011). Briefly, in the area, there are 373 municipalities: 60 in the state of Pará, 79 in the state of Tocantins, 186 in the state of Maranhão and 48 in the state of Piauí. Between 2001 and 2009, 82 463 new leprosy cases were notified in this cluster, with an annual mean detection rate of 95.9 per 100 000 population. In 15 year-olds, the annual mean detection rate was 28.4 cases per 100 000 (Alencar et al. 2011).

image

Figure 1.  Study area– cluster of high transmission risk situated in the states Maranhão, Pará, Tocantins and Piauí in north and northeast Brazil.

Download figure to PowerPoint

We present spatial analyses of the following indicators according to the Enhanced Global Strategy for Further Reducing the Disease Burden due to Leprosy by WHO (World Health Organization 2009): rate of new cases/100 000 population; rate of cases <15 years of age/100 000 population; rate of new cases with grade 2 disabilities (i.e. visible deformities)/100 000 population; and proportion of cases presenting with grade 2 disabilities. Spatial analysis was based on these indicators, with the municipalities as the observation units. Epidemiological data of municipalities were obtained from the four state databases for reported diseases (Sistema de Informação de Agravos de Notificação – SINAN). We included notifications from 1st January 2001 to December 31st 2009. The mean annual indicators of the entire observation period were used to smooth random variations in municipalities with fewer annual cases, as suggested previously (Gatrell & Bailey 1996). Population estimates of the respective years were obtained from the database of the Brazilian Institute of Geography and Statistics (Instituto Brasileiro de Geografia e Estatística, IBGE) (DATASUS 2011).

The database with the number of cases, estimation of the population size and indicators of the period was merged with census data from geographic databases. In a first step, descriptive thematic maps were produced for the selected indicators with ArcGis 9.3 software (Environmental Systems Research Institute, Redlands, CA, USA) (ESRI 2010). Then, we performed local empirical Bayesian analysis, through the software package TerraView version 3.6.0 (INPE, São José dos Campos, Brazil) to estimate smoothed indicators (TerraView 3.6.0 2010). We also used SaTScan software version 9.0.1 (Harvard Medical School, Boston, USA) to identify likely high-risk clusters (Kulldorff 2010b). This statistical technique uses a flexible geographic scanning window and includes different sets of neighbouring areas (Santos & Souza 2007; Kulldorff 2010a). The clusters were identified using pure spatial analysis, as proposed by Kulldorff and Nagarwalla (Kulldorff & Nagarwalla 1995; Kulldorff 1997), with a window radius of 100 km.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

In the cluster, 254 (68.0%) municipalities were classified as hyperendemic (mean annual detection rates >40 cases/100 000 inhabitants) with a maximum of 427.6 cases per 100 000 inhabitants. The distribution of crude detection rates is shown in Figure 2a. There was a concentration of municipalities with higher detection rates in Pará and in the center of Maranhão, totalling 52 (13.9%) municipalities with mean rates of >150 annual cases/100 000 inhabitants. In local Bayesian analysis (smoothed indicators), there was an increase in the number of hyperendemic municipalities (289; 77.4%) taking into account their closest neighbours and a small reduction in those over 150 cases per 100 000 inhabitants (13.4%) (Figure 2b).

image

Figure 2.  Spatial analysis in a highly endemic leprosy area using descriptive spatial statistics, Local Empirical Bayesian approach and Scan spatial statistics: mean annual rate of new cases/population (a–c); rate of new cases <15 year-olds (d–f); rate of new cases with grade-2 disabilities/population (g–i); proportion (%) of new cases with grade-2 disabilities (j, k).

Download figure to PowerPoint

Regarding detection in 15 year-olds, there were only 62 (16.6%) municipalities classified with low or medium endemicity (<2.5 cases/100 000 population under 15 years), and 219 (58.7%) municipalities classified as hyperendemic. Figure 2d shows the pattern of this indicator, with the majority of municipalities presenting high values in Pará and in the central areas of Maranhão. The smoothed coefficient more clearly identified these areas: there were 304 (81.5%) hyperendemic municipalities with 72 (23.6%) reaching smoothed values >40 cases/100 000 inhabitants (Figure 2e). The maximum value of this indicator was 100.3 cases/100 000 inhabitants.

The distribution of new cases with grade-2 disabilities per 100 000 inhabitants (Figure 2g) shows that 21.4% of municipalities did not have any leprosy grade-2 cases at diagnosis throughout the observation period, but 11.7% municipalities had >8 cases and up to 31 per 100 000 inhabitants. There was no clear pattern of this indicator in descriptive spatial analysis, but the smoothed indicator shows two defined areas, the first in the southeast of Pará and another in the center of Maranhão (Figure 2h). The smoothed maximum value of this indicator was 11.8/100 000 population.

The proportion of new cases with grade-2 disabilities highlights municipalities with high risk for late diagnosis mainly in southern Maranhão (Figure 2j). Only two (0.5%) municipalities in the cluster showed high values (>5%). Local Bayesian analysis highlights an increased proportion of new cases of leprosy with grade-2 (smoothed value of 2.3%) in south-eastern Maranhão (Figure 2k).

Spatial scan cluster analysis identified 23 likely clusters of increased new leprosy case detection rates, mostly in Pará and Maranhão (Table 1). These clusters included only 32% of the total population, but 55.4% of new leprosy cases in the cluster. The mean annual detection rate in 27.1% of these municipalities was 164 cases per 100 000 inhabitants (Table 1).

Table 1.   Statistically significant clusters of new cases of leprosy per 100 000 inhabitants defined using spatial scan statistics in a high endemic cluster in Brazil, 2001–2009
ClusterCentral municipality (state)N mun.Radius (km)Annual detectionRelative RiskP-value
 1Canaã dos Carajás (PA) 672.4247.82.65<0.0001
 2Novo Repartimento (PA) 476.5214.82.29<0.0001
 3Açailândia (MA)1573.5165.31.79<0.0001
 4Pio XII (MA)2575.2142.11.52<0.0001
 5Redenção (PA) 990.8194.22.04<0.0001
 6Paragominas (PA) 1 0189.71.98<0.0001
 7São João do Araguaia (PA)1164.3138.31.45<0.0001
 8Guaraí (TO) 1 0298.93.10<0.0001
 9Timon (MA) 1 0162.71.70<0.0001
10Itapecuru Mirim (MA) 1 0187.91.95<0.0001
11Tailândia (PA) 273.5158.91.65<0.0001
12Nova Olinda (TO) 554.5126.51.31<0.0001
13Amapá do Maranhão (MA) 221.3219.42.27<0.0001
14Floriano (PI) 1 0143.51.49<0.0001
15Pacajá (PA) 1 0148.51.54<0.0001
16Governador Archer (MA) 1 0201.42.08<0.0001
17Barrolândia (TO) 341.3152.61.58<0.0001
18Ourilândia do Norte (PA) 1 0152.01.57<0.0001
19Presidente Dutra (MA) 1 0133.31.38<0.0001
20Cristino Castro (PI) 1 0170.01.76<0.0001
21Tomé-Açu (PA) 1 0123.01.27<0.0001
22Maranhãozinho (MA) 525.0116.41.20=0.0013
23Itaipava do Grajaú (MA) 347.6116.71.21=0.012

Figure 2c shows the five most significant clusters for the general detection rate. Three of them are located in Pará, one in Maranhão, and one comprising Pará, Maranhão and Tocantins (Figure 2c). The most significant cluster includes six municipalities in the south of Pará (Relative Risk [RR] = 2.65) (Table 1).

We also identified 16 significant clusters for the detection rate <15 years (Table 2). The first three main clusters identified overlap with clusters of the general detection rate (Figure 2f). The fourth cluster of children <15 years of age (RR = 2.45) was equivalent to a cluster of newly detected cases. The fifth cluster has only five municipalities (RR = 2.61) in the center of Maranhão (Table 2). There were 11 other significant clusters, distributed throughout all four states, with a concentration in the southeast of Pará (Table 2).

Table 2.   Statistically significant clusters new cases of leprosy of children <15 years of age per 100 000 inhabitants defined using spatial scan statistics in a high endemic cluster in Brazil, 2001–2009
ClusterCentral municipality (state)N mun.Radius (km)Annual <15 detectionRelative RiskP-value
 1Canaã dos Carajás (PA) 672.483.02.77<0.0001
 2Breu Branco (PA) 562.269.62.42<0.0001
 3Açailândia (MA)1573.555.01.94<0.0001
 4Redenção (PA) 990.871.62.45<0.0001
 5Matões do Norte (MA) 534.476.42.61<0.0001
 6Pimdaré Mirim (MA) 27.7367.22.27<0.0001
 7São João do Araguaia (PA)1164.346.51.59<0.0001
 8Timom (MA) 1 051.51.74<0.0001
 9Paragominas (PA) 1 056.21.89<0.0001
10Bacabal (MA) 1 053.61.80<0.0001
11Guaraí (TO) 1 078.62.63<0.0001
12Floriano (PI) 1 054.91.84<0.001
13Arame (MA) 1 059.62.00<0.001
14Pacajá (PA) 1 054.51.83=0.014
15Amapá do Maranhão (MA) 221.367.42.25=0.067
16Trizidela do Vale (MA) 1 065.22.18=0.072

In addition, there were 11 likely clusters of new cases with grade-2 disability/population in Pará and Maranhão, with only one municipality in Tocantins (Table 3). The most significant cluster had a RR of 2.24 and an annual detection of 10.4 cases/100 000 inhabitants, in the center of Maranhão (Figure 2i). Clusters 2 and 3 overlap with new cases detection and detection of children under 15 years of age with leprosy. The proportion of new cases with grade-2 did not reveal any likely clusters.

Table 3.   Statistically significant clusters of leprosy cases with grade-2/100 000 inhabitants defined using spatial scan statistics in a high endemic cluster in Brazil, 2001–2009
ClusterCentral municipality (state)N mun.Radius (km)Annual grade-2/inh.Relative RiskP-value
 1São Luiz Gonzaga do Maranhão (MA)1143.810.42.42<0.0001
 2Açailândia (MA)1573.5 7.91.84<0.0001
 3Canaã dos Carajás (PA) 672.410.42.39<0.0001
 4Anajatuba (MA) 434.1 9.62.17<0.0001
 5Amapá do Maranhão (MA)1081.6 9.02.04<0.0001
 6Jacundá (PA) 467.6 8.61.95<0.0001
 7Miranorte (TO) 1 018.24.08<0.0001
 8Presidente Dutra (MA) 217.9 9.02.04<0.0001
 9São João dos Patos (MA) 1 013.53.04<0.0001
10Concórdia do Pará (PA) 1 012.92.89=0.0027
11Santa Luzia (MA)1060.2 6.51.47=0.0071

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Spatial analysis of health events aims to identify geographical patterns by means of risk maps of indicators, to point out areas of higher severity and to facilitate the planning of public health interventions (Assunção 2003; Castro & Singer 2007; Gauy et al. 2007; Santos & Souza 2007). Herein, we present besides descriptive spatial analysis of crude indicators, a spatial Bayesian approach and spatial scan statistics to describe in more detail leprosy disease dynamics in a highly priority extensive area in Brazil. Our population-based data show a heterogeneous geographical pattern of indicators, although not randomly distributed, and we identified well-defined spatial clusters of high risk for transmission and late diagnosis even inside a known leprosy risk area.

A problem associated with the use of crude indicators in spatial analysis is their high instability to express the risk of a certain event when this event is rare, and when the population of the region of occurrence is small (Gatrell & Bailey 1996; Assunção et al. 2005). In addition, municipalities with reduced leprosy case detection rates, which may be a result of political or operational factors, close to other municipalities with high detection rates may not reflect transmission dynamics adequately.

We used Bayesian inference that calculates uncertain values, which are not visible in a probabilistic way. The local empirical Bayesian estimator included spatial effects using only the closest neighbours of the area, converging towards a local mean rather than a global average (Gatrell & Bailey 1996; Assunção et al. 2005). The resulting smoothed indicators reduced these problems of variation between municipalities due to operational factors (Gatrell & Bailey 1996), and high-risk areas were identified more easily. In fact, the crude spatial distribution of new case detection rates indicates that there is probably gross influence of operational factors of the leprosy control programs in the four states involved: several municipalities with high detection rates bordered others with low detection. Local Bayesian method pointed out five areas with high detection rates, three in Pará, one in the center of Maranhão and another one encompassing municipalities of Pará, Maranhão and Tocantins. Historically, these areas show high values for detection rates; specifically, the largest detection of cases in the southeast of Pará coincides with the colonization in the 1970s, associated with the construction of a federal Amazonian highway, the BR-153 (Magalhães & Rojas 2007).

The use of Bayesian method mapping provides a useful addition to the analytical toolbox in the monitoring and surveillance of the geographical distribution of leprosy. This method provides a practical approach to the estimation of suspected under-registrations in a geographical area (Bailey et al. 2005). The supply and quality of leprosy health services is critical for the diagnosis and treatment of cases, and so, a poorly structured health service will have fewer cases, not due to a low endemicity, but the absence of active search for cases (Lapa et al. 2001).

Another way applied to identify vulnerable areas is the spatial scan statistic, which identifies most likely clusters of a possible event in a population with non-homogeneous spatial density, and simultaneously uses methods of statistical inference to test the significance of these clusters (Kulldorff & Nagarwalla 1995). We identified 23 likely clusters. The five most significant clusters are located in areas similar to those identified by the smoothed Bayesian coefficient, which strengthens evidence provided.

We made similar observations regarding detection rates in <15 year-olds. The detection rate in this subgroup is used as one of the main epidemiological indicators of leprosy control, as it expresses the strength of recent transmission and trends (Ministry of Health of Brazil 2008). It reflects the severity of the endemic level of leprosy and early exposure to Mycobacterium leprae (Alencar et al. 2008; Imbiriba et al. 2009). Spatial scan statistic identified 16 clusters of high transmission (i.e. detection rates in <15 year-olds). The seven most significant clusters include 53 municipalities. There are clusters covering more than 10 municipalities, such as cluster 3, with a detection rate in children <15 years of 55.0 cases per 100 000 inhabitants, covering areas of Maranhão, Pará and Tocantins.

The Brazilian Hansen’s Disease Control Program aims to identify these areas of high transmission, where populations are more exposed to concentrated sources of infection to reduce hidden prevalence and to interrupt the chain of disease transmission (Ministry of Health of Brazil 2008; Heukelbach et al. 2011). Almost all municipalities in Pará in the cluster had values >10 cases in <15 year-olds per 100 000 inhabitants. Identifying these areas at highest risk of disease will provide further evidence to the leprosy control program to direct control (Ministry of Health of Brazil 2008; Penna et al. 2009a,b).

Recently, an alternative indicator has been proposed by the World Health Organization to monitor leprosy control activities: the rate of grade-2 disabilities in new cases/population. This new indicator, when interpreted together with others, can be used to estimate under-detection, to measure the need for physical, social rehabilitation and delayed diagnosis, and to advocate activities for prevention of disabilities. Thus, this indicator should be prioritized for monitoring control measures (World Health Organization 2009; Oliveira et al. 2010; Alberts et al. 2011; Declercq 2011). In the cluster, there were 3811 cases diagnosed with grade-2 disability, a rate of 4.43 cases per 100 000 inhabitants (Alencar et al. 2011). The smoothed rates show clearly areas of higher risk: Pará and central Maranhão presented several municipalities with delayed leprosy detection, with rates up to nine times higher than the rest of Brazil. The identification of spatial clusters of grade-2/population again shows the similarity with the three areas identified in Figure 2c, 2f and 2i. As observed in the general detection rate and detection rate in children <15 years of age, the same municipalities that belong to the states of Pará, Maranhão and Tocantins also showed delayed diagnosis beyond intense transmission of leprosy. There was yet another area further south, fully in the state of Pará, with similar characteristics.

Another indicator for late diagnosis is the proportion of new cases presenting grade-2 disabilities. However, this indicator is strongly influenced by operational factors (Bakker et al. 2009). No significant clusters were identified with spatial scan statistics, indicating that this operational factor does not show considerable geographical clustering in our study area.

Several authors previously performed different approaches to analyse spatial patterns of leprosy. Martelli et al. (1995) showed that the geographical distribution of leprosy was far from uniform and described clearly defined high-risk areas in an ecologic study analysis in Goiania, central Brazil (Martelli et al. 1995). Another study conducted in Brazil’s Amazon region identified links between areas of intense deforestation and high rates of leprosy detection, highlighting not only environmental factors but also migration of possible importance for leprosy transmission dynamics (Silva et al. 2010). Our study identified clusters where the central municipalities are located within these same areas of deforestation. In a study of the spatial distribution of leprosy in the municipality of Mossoró in Rio Grande do Norte state, northeast Brazil, the use of Geographic Information Systems (GIS) allowed the identification of high-risk areas and significant clusters of leprosy inside a single municipality, showing a descriptive image of the spatial distribution of population indicators of leprosy (Dias et al. 2007).

Other studies include these approaches in other countries. In Indonesia, Bakker et al. (2004) identified significant clusters of leprosy patients varying from one to five houses. These clusters included 20 patients, of whom 13 were multibacillary patients, indicating the presence of active transmission in these areas. Spatial clustering of leprosy cases was also found on an island in Indonesia, with a particularly significant correlation in the population <21 and >39 years of age, and the spatial model could best explain the distribution of seropositive individuals (Bakker et al. 2004, 2005). In Bangladesh, a space-time analysis from 1989 to 2003 identified clusters and their dynamics over time, but was unable to explain the high incidence of leprosy in these clusters (Fischer et al. 2008). So far, there is no study available from Brazil or other endemic countries integrating the three spatial statistics methods on leprosy used in our study.

Our study has some limitations. Scan spatial analysis uses as geographical reference the coordinates of the urban centre, which may not reflect the distribution of cases inside the municipalities. Even with a high probability of the existence of a cluster, its accurate boundaries may not be so clear, and a more detailed local analysis of each municipality, as well as a temporal trend analysis of epidemiological and operational indicators leprosy may provide further evidence for these areas at greatest risk. In addition, clusters are always defined as circles or ellipses. An area of low frequency of cases surrounded by areas with the highest number of cases will be included in a cluster, although it may have different characteristics (Sankoh et al. 2001). Another limitation is the fact that in some cases, a single or few municipalities were classified as a cluster by itself due to the big size of the municipality’s territory, which was larger than the radius used to define the clusters. The use of secondary data may be another limitation, it often has inconsistencies, but may have been reduced because of the analysis used.

We conclude that spatial scan statistic is adequate to validate and confirm high-risk leprosy areas for transmission and late diagnosis, identified using descriptive spatial analysis and using local empirical Bayesian method. We identified municipality clusters, located inside a known leprosy high-risk area. Several indicators were used, assessing both health services and disease transmission dynamics. National and State leprosy control programs urgently need to intensify control actions in these highly vulnerable municipalities.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

This paper forms part of the MAPATOPI study, an interdisciplinary project providing evidence for improving the Brazilian Hansen’s Disease Control Program. The project is co-financed by the Brazilian Research Council (Conselho Nacional de Desenvolvimento Científico e Tecnológico, CNPq) and the Department of Science and Technology of the Brazilian Ministry of Health (DECIT). JH is research fellow from CNPq. We thank Fundação Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES, Brazil) for granting a PhD scholarship to CHA. The State Leprosy Control Programs of the State Health Secretariats of Maranhão, Pará, Tocantins and Piauí kindly provided data set.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Alberts CJ, Smith WCS, Meima A, Wang L & Richardus JH (2011) Potential effect of the World Health Organization’s 2011–2015 global leprosy strategy on the prevalence of grade 2 disability: a trend analysis. Bulletin of the World Health Organization 89, 487495.
  • Alencar CHM, Barbosa JC, Ramos AN Jr et al. (2008) Hanseníase no município de Fortaleza, CE, Brasil: aspectos epidemiológicos e operacionais em menores de 15 anos (1995–2006). Revista Brasileira de Enfermagem 61, 694700.
  • Alencar CH, Ramos AN Jr, Barbosa JC et al. (2011) Persisting leprosy transmission despite increased control measures in an endemic cluster in Brazil: the unfinished agenda. Leprosy Review. in press.
  • Assunção RM (2003) Estimadores Bayesianos Empíricos Espaciais de Taxas [Online]. Available: http://www.est.ufmg.br/~assuncao/cursos/espacial/aplicado/aulas/empbayes.pdf [Accessed 2011].
  • Assunção RM, Schmertmann CP, Potter JE & Cavenaghi SM (2005) Empirical Bayes estimation of demographic schedules for small areas. Demography 42, 537558.
  • Bailey TC, Carvalho MS, Lapa TM, Souza WV & Brewer MJ (2005) Modeling of under-detection of cases in disease surveillance. Annals of Epidemiology 15, 335343.
  • Bakker MI, Hatta M, Kwenang A et al. (2004) Population survey to determine risk factors for Mycobacterium leprae transmission and infection. International Journal of Epidemiology 33, 13291336.
  • Bakker MI, May L, Hatta M et al. (2005) Genetic, household and spatial clustering of leprosy on an island in Indonesia: a population-based study. BMC Medical Genetics 6, 40.
  • Bakker M, Scheelbeek P, Beers SM & Richardus J (2009) The use of GIS in leprosy control. Leprosy Review 80, 327331.
  • Castro MC & Singer BH (2007) Environment and health: a methodological approach for spatial assessment of malaria transmission in colonization projects in the Brazilian Amazon. Revista Brasileira de Estudos de Populacão 24, 247262.
  • DATASUS (2011) http://tabnet.datasus.gov.br/cgi/deftohtm.exe?ibge/cnv/popbr.def, [Accessed at 1/7/2011].
  • Declercq E (2011) Reflections on the new WHO leprosy indicator: the rate of new cases with grade 2 disabilities per 100 000 population per year. Leprosy Review 82, 35.
  • Dias MCS, Dias GH & Nobre ML (2007) The use of Geographical Information System (GIS) to improve active leprosy case finding campaigns in the municipality of Mossoro, Rio Grande do Norte State, Brazil. Leprosy Review 78, 261269.
  • ESRI (2010) ArcGis9.3, Environmental Systems Research Institute, Inc 1982–2011.
  • Feenstra SG, Nahar Q, Pahan D, Oskam L & Richardus JH (2011) Recent food shortage is associated with leprosy disease in bangladesh: a case-control study. PLoS Neglected Tropical Diseases 5, e1029.
  • Fischer E, Pahan D, Chowdhury S & Richardus J (2008) The spatial distribution of leprosy cases during 15 years of a leprosy control program in Bangladesh: an observational study. BMC Infectious Diseases 8, 126.
  • Gatrell A & Bailey T (1996) Interactive spatial data analysis in medical geography. Social Science and Medicine 42, 843855.
  • Gauy JS, Hino P & Santos CB (2007) Spatial distribution of leprosy cases in Ribeirao Preto, Brazil, 2004. Revista Latino-americana de Enfermagem 15, 460465.
  • Heukelbach J, Chichava AO, Oliveira AR et al. (2011) Interruption and defaulting of multidrug therapy against leprosy: population-based study in Brazil’s Savannah region. PLoS Neglected Tropical Diseases 5, e1031.
  • Holveck J, Ehrenberg J, Ault S et al. (2007) Prevention, control, and elimination of neglected diseases in the Americas: pathways to integrated, inter-programmatic, inter-sectoral action for health and development. BMC Public Health 7, 6.
  • Imbiriba ENB, Neto Silva AL, Souza WV et al. (2009) Desigualdade social, crescimento urbano e hanseníase em Manaus: abordagem espacial. Revista de Saude Publica 43, 656665.
  • Kerr-Pontes LR, Barreto ML, Evangelista CM et al. (2006) Socioeconomic, environmental, and behavioural risk factors for leprosy in North-east Brazil: results of a case-control study. International Journal of Epidemiology 35, 9941000.
  • Kulldorff M (1997) A spatial scan statistic. Comm Statist – Theory Meth 26, 14811496.
  • Kulldorff M (2010a) SaTScan User Guide for version 3.1. disponivel em http//http://www.satscan.org. Acessado em 12 de setembro de.
  • Kulldorff M (2010b) SaTScan v 9.0.1: Software for the Spatial and Space-Time Scan Statistics. Information Management Services Inc., Silver Spring, MD.
  • Kulldorff M & Nagarwalla N (1995) Spatial disease clusters: detection and inference. Statistics in Medicine 14, 799810.
  • Lapa T, Ximenes R, Silva NN et al. (2001) Vigilância da hanseníase em Olinda, Brasil, utilizando técnicas de análise espacial. Cadernos de saúde pública 17, 11531162.
  • Magalhães MCC & Rojas LI (2007) Spatial differentiation of leprosy in Brazil. Epidemiol Serv Saúde 16, 7584.
  • Martelli CM, Neto Moraes OL, Andrade AL et al. (1995) Spatial patterns of leprosy in an urban area of central Brazil. Bulletin of the World Health Organization 73, 315319.
  • Ministry of Health of Brazil (2008) Relatório Executivo do PNCH Ministério da Saúde. Período Maio de 2007 a Junho de 2008, Brasília, Secretaria de Vigilância em Saúde. Programa Nacional de Controle da Hanseníase. Mimeo.
  • Oliveira MLW, Grossi MAF, Oliveira CF et al. (2010) Commitment to reducing disability: the Brazilian experience. Leprosy Review 81, 342345.
  • Penna ML, Oliveira ML & Penna G (2009a) Spatial distribution of leprosy in the Amazon region of Brazil. Emerging Infectious Diseases 15, 650652.
  • Penna ML, Oliveira ML & Penna GO (2009b) The epidemiological behaviour of leprosy in Brazil. Leprosy Review 80, 332344.
  • Rodrigues LC & Lockwood DNJ (2011) Leprosy now: epidemiology, progress, challenges, and research gaps. Lancet Infect Diseases 11, 464470.
  • Sankoh OA, Ye Y, Sauerborn R, Muller O & Becher H (2001) Clustering of childhood mortality in rural Burkina Faso. International Journal of Epidemiology 30, 485492.
  • Santos SM & Souza WV (2007) Introdução á Estatística Espacial para a Saúde Pública, Brasília, Ministério da Saúde.
  • Silva DRX, Ignotti E, Souza-Santos R & Hacon SS (2010) Hanseníase, condições sociais e desmatamento na Amazônia brasileira. Revista Panamericana de Salud Publica 27, 268275.
  • TerraView 3.6.0. (2010). São José dos Campos, SP: INPE: 2010 [Online]. Available: http://www.dpi.inpe.br/terraview [Accessed 07/12/2010].
  • World Health Organization (2009) Enhanced Global Strategy for Further Reducing the Disease Burden due to Leprosy (Plan Period: 2011–2015). 28.
  • World Health Organization (2010) Global leprosy situation, 2009. Weekly Epidemiological Record, 85, 337348.