Rat infestation associated with environmental deficiencies in an urban slum community with high risk of leptospirosis transmission

We analyzed environmental factors that provide food, water and harborage to rodents and the risk of household rodent infestation in a slum community with a high risk of leptospirosis transmission. Detailed environmental surveys were performed in 221 households. Multivariate regression models evaluated the association between rodent infestation and socioeconomic status and environmental attributes obtained from Geographical Information System surveys. The general household infestation rate was 45.9%. Rattus norvegicus signs were the most prevalent, present in 74% of the infested households. The risk for rodent infestation was associated with environmental factors supporting harborage for rats, such as dilapidated fences/walls (OR: 8.95; 95%CI: 2.42-33.12) and households built on an earthen slope (OR: 4.68; 95%CI: 2.239.81). An increase of 1 meter from the nearest sewer was associated with a 3% (95%CI: 1%-5%) decrease in the risk of rodent infestation. A lack of sanitation where poor people live provides factors for rat infestation and could the target of educational interventions. Rodentia; Leptospirosis; Poverty Areas Correspondence N. J. Santos Escola de Medicina Veterinária, Universidade Federal da Bahia. Rua Barão de Jeremoabo s/n, Campus de Ondina, Salvador, BA 40170-115, Brasil. norlansantos@hotmail.com 1 Escola de Medicina Veterinária, Universidade Federal da Bahia, Salvador, Brasil. 2 Centro de Pesquisas Gonçalo Moniz, Fundação Oswaldo Cruz, Salvador, Brasil. 3 School of Public Health, Yale University, New Haven, U.S.A. 4 Instituto de Saúde Coletiva, Universidade Federal da Bahia, Salvador, Brasil. ARTIGO ARTICLE doi: 10.1590/0102-311X00132115 Cad. Saúde Pública 2017; 33(2):e00132115 This article is published in Open Access under the Creative Commons Attribution license, which allows use, distribution, and reproduction in any medium, without restrictions, as long as the original work is correctly cited.


Introduction
Leptospirosis, a disease caused by pathogenic spirochetes of the genus Leptospira, is acquired through contact with animal reservoirs or with a location contaminated by their urine 1 .In Brazil, approximately 10,000 leptospirosis cases are reported yearly with case fatality rates ranging between 10-15% 2 .Urban leptospirosis is an important health problem that has emerged due to recent growth of slums 3 .In these areas, leptospirosis affects communities where overcrowding, poverty, and the lack of basic sanitation services create conditions supporting leptospirosis transmission 4,5,6 .Additionally, those areas are at risk of other rodent-borne transmission given that rats are carriers of Seoul Virus, Bartonella and other pathogens.Although several animal species may carry Leptospira, synanthropic rodents, from the genus Rattus, are recognized as the most important reservoirs in urban areas 7,8,9 .Prevention of leptospirosis is essential given the challenges of leptospirosis diagnosis and clinical treatment as well as the lack of a human vaccine.Campaigns for rodent control, based on the use of chemical rodenticides, have been a major strategy for preventing urban leptospirosis 10 .However, chemical interventions are not effective for long-term control of rodent infestations 11 and have important limitations such as the development of rodenticide resistance and potential adverse impacts on non-target species 12 .A few major cities in Brazil (São Paulo, Salvador, Recife and Curitiba) have shifted the focus of their interventions from rodenticide-focused pest control to a more holistic approach of integrated interventions (e.g., education, poisoning and environmental modifications).Nevertheless, the lack of knowledge regarding the social and ecological determinants influencing rodent infestation in these settings has been a barrier to implement more effective rodent control 13,14,15,16 .
Few studies have examined the environmental determinants of rat infestation in urban habitats in developing countries 17,18 .In Brazil, studies to determine factors associated with rodent infestation have been restricted to São Paulo 19 in areas of low leptospirosis incidence.Characterizing the determinants of reservoir infestation is critical to improve leptospirosis prevention.In this work we analyzed the environmental factors that provide food, water, and harborage and the risk of household rodent infestation in a community with high risk of leptospirosis transmission.

Study site and study design
The study area, Pau da Lima (Figure 1), is situated in the outskirts of Salvador, a city of 2,892,625 inhabitants 20 located in Northeast Brazil.Pau da Lima is a densely-populated community of 0.46km 2 divided into four valleys.The area has a high incidence of Leptospira transmission.Incidence of asymptomatic Leptospira infection was 3,780 per 100,000 inhabitants, as determined in cohort studies.Moreover, active hospital-based surveillance identified an incidence rate of 35.4 severe cases of leptospirosis per 100,000 inhabitants 4,5 .Between 2003 and 2007 a nested case control study was performed to identify environmental variables in households with and without evidence of Leptospira transmission 4 .We subsequently used a convenience sample of previously collected environmental survey data 4 to assess environmental variables associated with rodent infestation.
Ethical clearance for this study was granted by the Ethics Research Committee of the Oswaldo Cruz Foundation and the Institutional Review Board of Weill Cornell Medical College at Cornell University.

Data collection
A household environmental survey was performed during October and November of 2007, as described previously 4 .Briefly, we adapted and used a form from the US Centers for Disease Control and Prevention (CDC) manual 13 .The form included the following six groups of variables:  for rodents; and (f) six variables on signs of rodent infestation.The presence of active signs of rodents qualified as evidence of rodent infestation.
We combined households with and without evidence of Leptospira transmission to evaluate the rodent infestation rate (RIR).Infestation was estimated as follows: GIR= [(P1 * N1) + (P2 * N2)] / T where: GIR (general infestation rate), P1 (proportion of infested households with Leptospira transmission), N1 (number of households in Pau da Lima with Leptospira transmission), P2 (proportion of infested households without Leptospira transmission), N2 (number of control households in Pau da Lima without Leptospira transmission) and T (total number of households in Pau da Lima).
Geographic Information Systems (GIS) were used to measure the distance to environmental variables (Figure 2), such as the location and size of open refuse deposits, open sewage, and rainwater drainage systems 6 .

Statistical analysis
Data for individual subjects were linked by location of residence to spatially-coded information of households and environmental attributes within the study site.Chi-square and t-test were used to compare categorical and continuous data, respectively, in the bivariate analysis.
Generalized additive models (GAM) 21 were used to evaluate the association between significant continuous variables and the rodent infestation.Variables with p < 0.1 in the bivariate analysis were then included in a multivariate, logistic regression analysis.Seven models were used to fit the different blocks of variables based on different types of determinants of rodent infestation.The first model included socioeconomic variables.Models two through six included environmental variables that we hypothesized were related to rodent infestation.The seventh and final model included retained variables from the previous six multivariate models.A backward elimination strategy was performed to obtain models using SAS for Windows software (SAS Inst., Cary, USA).A p-value of 0.05 or less was used as criteria for a statistically significant difference.

Table 1
Active rodent signs identified in 137 infested households.

Active rodent signs
Infested households n %

Results
We evaluated 221 households in the study area, from a total of 684; of these, there was evidence of Leptospira infection in 80; and no evidence of infection in 141.Based on the environmental survey, 137 of the 221 households (62%) presented at least one sign of rodent active.In the households with evidence of Leptospira infection, 63 (78%) had signs of rodent infestation.In addition, 74 (52%) of the households with no evidence of Leptospira transmission were infested.Considering the total number of households with evidence of Leptospira infection (n = 80) and control households (n = 604), the general infestation rate for the study area was calculated at 55%.More prevalent active rodent signs identified between evaluated households were rodent burrows (79%), rodent runs (76%) and R. norvegicus feces (74%).Feces of Mus musculus and Rattus rattus were less frequent (Table 1).The overall rodent infestation distribution can be seen in Figure 3.
In bivariate analyses, we found that the number of household inhabitants and per capita income were associated with the occurrence of rat infestation in the household (Table 2).From the environmental variables 24 were associated with rodent infestation: five such variables related to house characteristics, seven related to water or food for rodents, eight related to harborage, and four to access to the house.Additionally, the risk of rodent infestation was not associated with dogs or cats in the household (Table 2).GAM analysis showed that the risk of rodent infestation had an inverse linear association with the distance of the household to an open sewer (Figure 4).Additionally risk of rodent infestation had an inverse association with increasing elevation of the household from the lowest point in the valley (Figure 5).Increased risk was observed among households located less than a threshold distance of 25 meters from an open sewer or the lowest point of the valley.Of note, the distance of the household to an open sewer was highly correlated with household elevation (Pearson's correlation coefficient = 0.61; p < 0.001) since open sewers drain into the bottoms of valleys.
To identify independent predictor variables for rodent infestation, we performed multivariate analysis (Table 3).In the seventh and final model, we retained the following five variables: one variable from the premise details group, three variables from the harborage of rodents group and one from the access to rodents group.The risk for rodent infestation was associated with harborage, water, food for rodents, and a quality house indicator.Inappropriate organic refuse storage was associated with approximately a four-time greater risk (OR: 3.98; 95%CI: 2.12-7.46).Dilapidated fences, household built on exposed earth slope, rubbish, and bush or shrubbery were significant rodent harborageassociated risk factors.Households located less than 25 meters from an open sewer and at the lowest point in the valley had a 2.8 times (95%CI: 1.3-6.0)increased risk of rodent infestation compared with those located 25 meters or more from these attributes.Un-plastered exterior wall surface and holes in the roof were significant risk factors to rodent access to the household.

Discussion
The results of our household surveys indicate that rats were widespread in the study area of Pau da Lima during the sampling period in 2007.A high degree of infestation, reaching almost 50% of households, is not surprising in light of the absence of any significant rodent control measures in Pau da Lima during the survey period.It also reflects the general trend seen in urban areas around the world 19,22 , where the poorer the conditions, the higher the level of infestation 23 .Areas with infestation rates above 25% are at high risk of rodent borne diseases 13    identified R. norvegicus as the dominant species in the urban ecosystem of Salvador 4 , brown rat feces represented 80% of the rodent feces in our study.This pattern is also observed in other urban areas around the world 22,24,25,26,27,28,29,30 .
The presence of open sewers, which was retained in the model, was an independent risk factor for rodent infestation, likely as the source of harborage.It has been shown that R. norvegicus prefers environments with available water, and their presence has been associated with sewers 11,19,31 .In addition, GAM analysis found that households had a higher risk of rat infestation when situated within 25 meters from an open sewer (Figure 4) or at the lowest point in the valley (Figure 5).The finding that households had increased infestation risk when located within 25 meters from the lowest point in the valley or sewer suggests that this distance is a proxy for the degree of rat colonization from potential source habitats (sewer) in peri-domiciliary environments.
We observed that food, harborage, and entry points were highly accessible to rodents in the neighborhoods studied.In these complex and saturated habitats, we identified five independent risk factors for rodent infestation.Four of them belong to the harborage category, indicating that suitable places for hiding and nesting are key risk factors to rat infestation in urban ecosystems.In the present study the location of the households on an earthen slope was identified as an independent risk factor for infestation.Similar results were found in previous studies 24,29,31 .Households built on an incline provide suitable space for rodent harborage, and the preference of R. norvegicus to select slopes to build burrows was registered in a previous study 32 .The presence of dilapidated fences and walls are described as risk factors for rats in rodent control manuals of both the Brazilian Ministry of Health 10 and the CDC 13 .The presence of a dilapidated fence was a risk factor for R. rattus but not for R. norvegicus in São Paulo 19 .Houses with lumber/clutter on the ground had almost nine times the risk of infestation compared to households without this feature.Fences and walls in disrepair provide rodent harborage, which is also associated with rodent infestation in Brazil 19 .

Figure 5
Generalized additive models (GAM) of the association between the risk of rodent infestation and continuous variables of level of household in meters above the lowest point in valley.
Note: the coefficient F (infestation) in the GAM model is a measure for the risk of rodent infestation.In the panels the x axis intercept values, where F (infestation) equals zero and were the distance of household above the lowest point in the valley is 25 meters.
We observed a large availability of access points for rodents in the neighborhood studied.The only independent risk factor belonging to this group was holes in the roof, a variable initially proposed for rodent access.We believe that holes in the roof may be a proxy for socioeconomic status not captured by the income variable, because R. norvegicus, the rodent most frequently found, is not considered a good climber 33 .
The main limitation of the study was the selection for convenience which makes it difficult to generalize our results.In addition the study area has a complex structure with the existence of vegetation, garbage and objects that may obstruct the view of signs of infestation and thus result in an underestimation of infestation rates.Finally, the environmental assessment was carried out in a single season (dry) and so we would encourage further investigation in another season (rainy) due to possible seasonal changes in populations of R. norvegicus as observed in temperate countries 4 .

Conclusion
In conclusion, our findings showed that a high proportion (> 44%) of urban slum households are infested with R. norvegicus.The presence of a predominant rodent reservoir in a slum habitat with high risk of leptospirosis transmission highlights the importance of performing an integrated pest management (IPM) program based on R. norvegicus ecologies.Intervention must focus on basic sanitation and the elimination of rodent harborage sources, the major causes of rodent infestation in this Cad.Saúde Pública 2017; 33(2):e00132115 (a) seven variables on premise type; (b) five variables on food sources for rodents; (c) three variables on water sources for rodents; (d) eleven variables on harborage for rodents; (e) five variables on entry/access Cad.Saúde Pública 2017; 33(2):e00132115

Figure 1
Figure 1Slum community site in the city of Salvador, Bahia State, Brazil, and the aerial photograph is the boundary of the study site.

Figure 2
Figure 2Topographic map generated by the digital terrain model.

Figure 3
Figure 3Spatial distribution of rodent infestation.

Figure 4
Figure 4Generalized additive models (GAM) of the association between the risk of rodent infestation and continuous variables of level of household in meters and the distance in meters to the nearest open sewer.

Table 2
Environmental characteristics associated with rat infestation at the community study site.Salvador, Bahia State, Brazil.
* Variable definitions available in a previous study 4 ; ** Numbers and percentages are shown for categorical variables.Median and interquartile range (IQR) are shown for continuous variables of per capita household income, number of persons in the household and presence of an open sewer.*** p-value by chi square test was estimated for categorical variables.T-test was used for continuous data (continuous data can be identified by the presence of IQR).Cad.Saúde Pública 2017; 33(2):e00132115

Table 3
Risk factors for rat infestation.Salvador, Bahia State, Brazil.The presence of lumber and dilapidated fences is associated with human behavior and could be targets for educational interventions.Also it is important to maintain constant monitoring of populations of R. norvegicus by environmental and health professionals.These results can be used in developing rodent management programs to predict individual household probabilities for rat infestation, to direct control measures, and to assist policymakers in allocating limited resources.Santos and F. Costa participated in the project, data analysis and interpretation, as well as a significant contribution towards writing the article.They carried out relevant critical revision of the intellectual content of the article and approved the final version for publication.E.Sousa, M. G. Reis, and A. I. Ko contributed in writing up the article and approval of the final version for publication.El estudio analizó factores ambientales que facilitan la presencia de alimento, agua y abrigo para roedores y el riesgo de infestación por roedores en una comunidad sin recursos con alto riesgo de transmisión de la leptospirosis.Se realizaron encuestas ambientales detalladas en 221 domicilios.Los modelos de regresión multivariada evaluaron la asociación entre infestación por roedores, nivel socioeconómico y características ambientales, obtenidos a través de encuestas con Sistemas de Información Geográfica.El estudio mostró una tasa global de un 45,9% de infestación domiciliaria.Los indicios de Rattus norvegicus eran los más prevalentes, presentes en un 74% de los domicilios infestados.El riesgo de infestación por roedores estuvo asociado a factores ambientales que proporcionaban abrigo a las ratas, tales como cercas y muros derruidos (OR: 8,95; IC95%: 2,42-33,12) y domicilios construidos directamente sobre pendientes (OR: 4,68; IC95%: 2,23-9,81).Cada incremento de un metro de distancia, a partir del alcantarillado más próximo, estuvo asociado a una disminución de un 3% (IC95%: 1%-5%) en el riesgo de infestación por roedores.La falta de saneamiento básico en las viviendas de las familias pobres facilita la infestación por ratas y es el objetivo prioritario para intervenciones educativas.