Dengue

Dengue Bulletin Volume 28 (2004)

 PDF Version

Ecological Study of Rio de JaneiroCity  DEN-3 Epidemic, 2001-2002

Maria Lucia F. Penna^#

Escola Nacional de Saude Publica, Fundação Oswaldo Cruz, Rua Leopoldo Bulhões 1480,
DENSP, 21041-210 Rio de Janeiro, RJ, Brazil

Introduction

After decades of freedom from dengue virus infection, Brazil experienced an outbreak of DEN-1 and DEN-4 in areas close to the borders with Venezuela in 1981-82. Intensive vector control measures successfully controlled this outbreak and checked its spread to other areas in the country^[1]. In 1986, DEN-1 was introduced in the Rio de Janeiro metropolitan area, which spread to other areas, causing a massive epidemic that later covered the entire country. DEN-2 and DEN-3 were also
introduced in the country in 1990 and 2001, respectively^[2]. Once introduced in the Rio de Janeiro metropolitan area, DEN-3 caused a large epidemic during 2001 and 2002. In 2003, DEN-1, DEN-2 and DEN-3 were in circulation in all except the two southern states of the country, with 341,092 cases reported (Figure 1)^[3].

Despite the use of a variety of control strategies, dengue control is a major public health challenge in the world. Demographic changes occurring in developing countries due to widespread rural-urban migration epidemic that later covered the entire country. DEN-2 and DEN-3 were also
introduced in the country in 1990 and 2001, respectively^[2]. Once introduced in the Rio de Janeiro metropolitan area, DEN-3 caused a large epidemic during 2001 and 2002. In 2003, DEN-1, DEN-2 and DEN-3 were in circulation in all except the two southern states of the country, with 341,092 cases reported (Figure 1)^[3].

Despite the use of a variety of control strategies, dengue control is a major public health challenge in the world. Demographic changes occurring in developing countries due to widespread rural-urban migration
since the 1960s have resulted in overcrowded cities with multiple deficiencies, particularly in housing and basic sanitation. The strategy that resulted in the eradication of Aedes aegypti in the 70s is no longer applicable to the reality of the social, demographic, economic and political situation in South American countries^[4-6].

Figure. Circulation of dengue virus in the Brazilian states, 2003

None   DEN-1 and -3   DEN-1 and -2   DEN-1, -2 and -3

Predicting the risk of dengue correctly based on sociocultural factors has been the goal of many authors^[7,8], but the dynamics of dengue transmission in urban settings are still poorly understood^[9]. The understanding of the virus transmission dynamics requires a theoretical framework that includes individuals, households and behavioural risk factors as well as the administrative aspects of city management services, use of city space and movement of people across the metropolis. This understanding can help control measures to be more effective with responsibilities divided between the government and communities as per priorities established on the basis of reliable data.

The present study looked into the relationship between the urban socioeconomic organization and the dengue attack rate during the DEN-3 epidemic in Rio de Janeiro city in 2001-2002. It is mainly an exploratory study, based on available secondary data, that is likely to throw some light on how urban settings contribute to dengue transmission.

Methods

Description of study area

Rio de Janeiro city is located at -22°54'23" south latitude and -43°10'21" west longitude, at the seashore, with an urban area of 1,255 km², including inland and continental waters. The municipal area was divided into 26 administrative regions (ARs) in 1999 which were used as spatial units for analysis. The climate is tropical, hot and humid, with local variations due to altitude differences, vegetation and proximity to the sea. The mean annual temperature is 22 °C, with high daily means in summer (30 to 32 °C). Rainfall is 1,200 to 1,800 mm per year, concentrated in summer from December to March. The city has the lowest rate of population growth among the Brazilian capital cities, 6.9% between 1991 and 2000. 

Database

This study used secondary data published in the website of the city administration^[10], including the proportion of households belonging tothe sewage collection system (SEWAGE), the proportion of households belonging to the water supply system (WATER), the proportion of households with a well (WELL), the proportion of households with waste collection by the city authority (WASTE), the number of inhabitants per household (INH/HOUSE), the area of residential properties (square metres) per inhabitant (M²/INH), the proportion of households in slums (SLUMS), the proportion of the total area built for residence (HOUSE), for industry (INDUSTRY), for commerce or service (COMMERCE); the proportion of unused areas (UNUSED); the proportion of the total area of parks and squares (PARKS); the amount of city tax collected per inhabitant (TAX); the proportion of city tax of commercial origin (COMTAX), industrial origin (INDTAX) and service origin (SERVTAX); the mean per capita income (INCOME); the life expectancy (LE); the proportion of literacy among inhabitants (LITERACY); and the proportion of children aged 7 to 15 years attending school (SCHOOL) as independent variables. All proportions were presented as percentages and income and taxes as 1,000 reais (Brazilian currency). The independent variable was the attack rate for DEN-3 epidemic, calculated by the number of reported dengue cases during 2001-2002 divided by the population estimate for January 2001, per 100,000 inhabitants.

Statistical methods

A multiple regression model was adjusted to the data using stepwise forward approach, with F to enter = 1 and F to leave = 0.95, and the author interference to limit the number of steps in order to prevent a saturated model. The independent variable was given a log transformation. The software used was Statistica, from Statsoft^[11].

Results

Table 1 shows the correlation coefficients between the independent variable and all those variables presented in the model, along with their range. The model included as predictors the proportion of households with a well, the proportion of the available area used for commerce and services, the proportion of city taxes collected from industries, the mean per capita income and the built household area per inhabitant (Table 2), all with a statistical significant level less than 0.05. The model explains 71% of the attack rate variance. Table 3 shows the partial correlation coefficients for the variables included in the model, which is a measure of the association of each variable when the other variables are controlled for, making it possible to rank their effect. The residues fitted well into a normal distribution and presented no correlation with the predictors. 

Table 1. Descriptive statistics and correlation coefficient with log (attack rate)

Table 2. Regression summary
(Standard regression coefficient, regression coefficient, t and P value)

R= 0.84282066; R²=0.71034667; F(5,20)=9.8096; P<0.00007; Standard error of estimate=0.35169

Table 3. Partial and semi-partial correlation

Discussion

The exploratory aspect of this study justifies the presentation of a high number of urban and socioeconomic variables in the model. These variables present high covariance that induces the use of the forward stepwise method. To avoid a saturated model, the author restrained the number of steps and assured that only significant variables were kept in the model by using a relatively high value of F to leave. The rupture of the homocedasticy assumption due to the nature of data (proportions) was successfully dealt with by the logarithmic transformation of the attack rate^[12], as shown by the residual analysis.

The proportion of households with a well is really a proxy variable of the discontinuity of water supply. The proportion of households that have access to the city’s water supply network is 95.07% for the entire city, indicating a high coverage of the city’s water supply network. But the water supply is not equally effective in all the administrative regions (ARs), with some areas having chronic problems, mainly discontinuous supply. The presence of a well, although small, only in 1.12% of the households (maximum of 11.85% in PaquetaAR and 11.06% in GuaratibaAR) is a proxy variable for the discontinuity problem, for it is a solution that involves financial expenses which is only justified in the face of an important and lasting problem of supply. These results point to the fact that water supply is an important issue in dengue control, but only in the context of lack of or irregular water supply where the population is forced to resort to storage, which creates breeding sites for the vector and not in the usual context of adequate supply as suggested by other authors^[13].^.

The model included mean per capita income as a protective factor meaning that low socioeconomic status of residents of an AR is a risk factor for dengue transmission. A study in Brazil found no association between the socioeconomic status and dengue risk at the individual level^[14] and suggested that the previous finding of such an association at the aggregate level^[15,16] was a fallacy. However, it should be noted that studies that focus on factors at individual level are insufficient to address the ecological links in the causal chain. Ecological studies may be, as pointed out by Koopman and Languini^[17], the only way to study risk factors for infectious diseases. The risk of dengue virus infection is not dependent on the physiological characteristics of individuals, but on the environmental characteristics of the area where the group of individuals live. This includes other individuals, the natural environment and the way it is transformed by humans. The mean per capita income has to be interpreted as an environmental measure concerning the area and not as an aggregate measure of individuals, because it impacts the neighbouring households as well as the nearby public areas, thus affecting the presence of potential breeding sites for the vector. This discussion reinforces the relevance of multi-level analysis in the evaluation of dengue risk factors.

The inclusion in the model of the proportion of the area used for commerce and services and the proportion of city taxes paid by industry shows that urban organization plays an important role in the distribution of dengue. These two variables are proxy variables for the movement of people across ARs. The presence of commerce and general services was represented by the proportion of the area and the presence of industry as the proportion of city taxes, because those are the variables correlated to the intensity of the movement of people. The luxury commerce and services may collect more taxes than the popular commerce and services, but the areas of popular commerce and services have a much bigger flow of people. On the other hand, the relative size of the industrial area is related to the type of industry, as for instance big industrial storage areas, and not to the production or the number of workers. Industries that collect more taxes have a larger production and are more likely to have a higher number of workers. Other authors^[18] indicate that the probability of being reached by a new dengue virus is correlated to the intensity of communication among people and the density of traffic and the road network. It is important to note that the diffusion of the epidemic by proximity may be less important than the diffusion caused by the circulation of people in central areas in big cities, which is supported by the present finding. This area initially imports infected individuals from the initial foci of the virus where it is newly introduced, resulting in higher local transmission, which, in turn, results in exportation of infected individuals to other areas, reinforcing the epidemic all over the city. This fact clearly indicates the priority of mosquito control in areas with high levels of population mobility, such as the Rio de Janeiro city central area.

The model also included the area of residential property per inhabitant as a risk factor for dengue. This variable is closely correlated with the socioeconomic status represented by the mean per capita income (R=0.864605) that is controlled in the model, meaning that, for the same socioeconomic status, a bigger area of residential property per inhabitant results in bigger dengue risk. This is possibly due to the existence of empty spaces in properties, such as backyards, thus creating greater opportunities for the existence of the vector breeding sites. This hypothesis has to be investigated for its importance in the selection of priorities for vector control in private residences as well as in defining the contents of educational interventions.

The results of the present study allow the establishment of priorities in vector control and educational interventions, highlighting the importance of the flow of people across cities and its significance in dengue epidemics.

Conclusion

The current efforts to control dengue demand a more comprehensive approach, including health education, community participation, garbage disposal and proper urban planning, besides chemical and biological mosquito control measures. In order to improve the efficiency of control efforts, these activities have to concentrate on areas and populations at higher risk, which implies early identification of higher incidence periods and areas and their characteristics^[5].

In Rio de Janeiro, emphasis has been placed on bringing about behavioural changes in the communities to make an impact on the determinants and risk factors of dengue through educational interventions. Health authorities and the press attributed the main responsibility of the problem to public behaviour, not clearly distinguishing between the responsibilities of the government and that of the private citizen^[6].

References

1.      Osanai CH, Travassos-da-Rosa APA, Amaral S, Passos ACD and Tauil PL. Surto de dengue em Boa Vista, Roraima. Revista do Instituto de Medicina Tropical de Sao Paulo, 1983, 1: 53-54.

2.      Nogueira RM, Miagostovich MP, Schatzmayr HG, dos Santos FB, de Araujo ES, de Filippis AM, de Souza RV, Zagne SM, Nicolai C, Baran M and Teixeira Filho G. Dengue in the State of Rio de Janeiro, Brazil, 1986-1998. Mem Inst Oswaldo Cruz, 1999, 94(3): 297-304.

3.      Secretaria de Vigilância em Saúde. Dengue Boletim semana 16; 2004 online http://dtr2001.saude.gov.br/svs/epi/dengue/boletim/pdfs/Boletim%20semana%2016%202004.pdf.

4.      Tauil PL. Urbanização e ecologia do dengue. Cad Saúde Pública, 2001, 17(1): 99-102.

5.      Pan American Health Organization. A blueprint for action for the next generation: dengue prevention and control. Washington: PAHO, 1999, 1-14.

6.      Penna MLP. Dengue control: a challenge for the public health system in Brazil. Cad Saúde Pública, 2003, 19(1): 305-309.

7.      Hayes JM, Garcia Rivera, Flores Reyna, Suarez Rangel, Rodríguez Mata, Coto Portillo, Baltrons Orellana, Mendoza Rodríguez, De Garay, Jubis Estrada, Hernández Argueta, Biggerstaff BJ and Rigau Perez. Risk factors for infection during a severe dengue outbreak in El Salvador in 2000. Am J Trop Med Hyg, 2003, 69: 629-633.

8.      Ashford DA, Savage HM, Hajjeh RA, McReady J, Bartholomew DM, Spiegel RA, Vorndam V, Clark GG, Gubler DG. Outbreak of dengue fever in Palau, Western Pacific: risk factors for infection. Am J Trop Med Hyg, 2003, 69: 135-140.

9.      C‚mara tecnica e cientifica de assessoramento ao controle do dengue. Recomendações à Secretaria de Estado de Saude. Secretaria de Estado de Saúde do Rio de Janeiro, Rio de Janeiro, RJ, 2002, 1-11.

10.   Instituto Municipal de Urbanismo Pereira Passos, Secretaria Municipal de Urbanismo, Cidade do Rio de Janeiro. Website: http://www.armazemdedados.rio.rj.gov.br/, April 2004.

11.   StatSoft Inc. Electronic Statistics Textbook. Tulsa, OK: StatSoft. http://www.statsoft.com/ textbook/stathome.html, 2004.

12.   Selvin S. Epidemiologic analysis: a case-oriented approach. Oxford University Press, New York, 2001: 321.

13.   Forattini Oswaldo Paulo and e Brito Marylene de. Reservatórios domiciliares de água e controle do Aedes aegypti. Rev Saúde Pública, 2003, 37: 676-677.

14.   Teixeira M, Barreto ML, Costa M, Ferreira LD, Vasconcelos PF and Cairncross S. Dynamics of dengue virus circulation: a silent epidemic in a complex urban area. Trop Med Int Health, 2002, 7: 757-762.

15.   Vasconcelos PF, de Menezes, Melo LP, Pesso ET, Rodríguez SG, da Rosa, Timbo MJ, Coelho IC and Montenegro F. A large epidemic of dengue fever with dengue hemorrhagic cases in Ceará State, Brazil, 1994. Rev Inst Med Trop Sao Paulo, 37: 253-255.

16.   Figueiredo LT, Cavalcante SM and Simoes MC. Dengue serologic survey of schoolchildren in Rio de Janeiro, Brazil, in 1986 and 1987. Bull Pan Am Health Organ, 1990, 24: 217-225.

17.   Longini IM Jr, Koopman JS, Haber M and Cotsonis GA. Statistical inference for infectious diseases. Risk-specific household and community transmission parameters. Am J Epidemiol, 1988, 128(4): 845-859.

18.   Muttitanon W, Kongthong P, Kongkanon C, Yoksan S, Gonzalez JP and Barbazan P. Spatial and temporal dynamics of dengue haemorrhagic fever epidemics In: Conference Proceedings of Map Asia 2002, GIS Development, 2002, http://www.gisdevelopment.net/application/health/planning/healthp0010c.htm