Dengue dynamics under climate drivers: analysis with ecological and epidemiological frameworks
Carregando...
Arquivos
Data
2022-10-31
Autores
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
A dengue tem sido uma doença constante nos cenários ecológico e epidemiológico brasileiros
desde os meados do século XX. A partir da década de 1990 os casos que, antes, eram
esporádicos, e se distribuíam sem qualquer clara associação a variáveis populacionais e do
território, começaram a ter uma nova dinâmica. Em 2001, por uma lei que tornou obrigatória a
notificação de casos de dengue, as notificações da doença explodem e a dengue passa a ser
vista como um problema de saúde pública.
A dengue é uma doença transmitida pela picada do mosquito Aedes Aegypti. Ela tem uma
componente sazonal. Porém, ainda que se entenda que essa sazonalidade seja derivada do
clima, ainda nos faltam melhores análises quantitativas dos efeitos do clima sobre as
epidemias de dengue.
Nesta tese, as séries temporais de casos de dengue e de variáveis climáticas essenciais (VCE),
como temperatura e precipitação, são analisadas através do mapeamento cruzado convergente
(CCM). O CCM se baseia no resultado do teorema de Takens que diz que o atrator de um
sistema dinâmico pode ser reconstruído a partir das séries temporais de uma das variáveis do
sistema, e as séries temporais desta mesma variável com atrasos temporais. A partir deste
resultado desenvolvem-se métodos de determinação de relações causais. Utilizamos este
método e mostramos que, no caso da cidade do Rio de Janeiro, a precipitação é a VCE mais
importante em relação a epidemias de dengue.
A dengue, de forma menos comum, pode levar a uma hospitalização conforme histórico de
infecção anterior e devido às condições climáticas. Lançando mão das séries-temporais de
casos de dengue hospitalizados e de séries de temperatura, exploramos quais são os riscos de
se desenvolver um quadro mais grave de dengue dada a exposição à temperatura. Para tal
utilizamos modelos não-lineares de atrasos distribuídos (DLNM), que ao colocar atrasos de
uma série de um fator de exposição num modelo generalizado linear com preditores da série
resposta, no nosso caso a série de hospitalizações por dengue, fornece uma associação
estatística entre esses fatores com a qual podemos contabilizar riscos relativos para cada
referencial de temperatura. Em um desenvolvimento posterior e com todas as associações
feitas ao nível municipal, fazemos uma meta-análise dessa associação, primeiro por macro
regiões do Brasil e segundo para o país todo, do qual retiramos uma forma funcional para o
risco relativo de cada percentil de temperatura tem em tornar um caso de dengue uma
possível hospitalização por dengue
Dengue has been a disease present in the Brazilian ecological and epidemiological scenarios since the mid-twentieth century. From the 1990s, the cases that, before, were sporadic,and distributed without any clear association to population and territory variables, began to exhibit a new dynamic. In 2001, by a law that made notification of dengue cases mandatory, notifications of the disease exploded and dengue began to be seen as a public health problem. Dengue is a disease transmitted by the mosquito Aedes Aegypti. It has a clear seasonal component. However, even if it is understood that certain seasonality is derived from climate, we still lack better quantitative analyses of the effects of climate on dengue epidemics. In this thesis, time-series of dengue cases and essential climate variables (ECV), such as temperature and precipitation, are analysed using convergent cross-mapping (CCM). CCM is based on the result of the Takens theorem which states that the attractor of a dynamical system can be reconstructed from the time-series of one of the variables of the system, and the time series of this same variable with time delays. Based on this result, methods for determining causal relations have been developed. We use these methods and show that, in the case of the city of Rio de Janeiro, precipitation is the most important ECV relent for dengue epidemics.. Dengue, in a less common way, can lead to hospitalization according to previous infection history and weather conditions. Using time-series of hospitalized dengue cases and temperature series, we explore which are the risks of developing a more severe dengue condition given temperature exposure. To this end, we use distributed lag non-linear models (DLNM), which, by using delays of a series of an exposure factor in a generalized linear model with predictors of the response series, in our case the series of dengue hospitalizations, provides a statistical association between these factors. This allows us to account for relative risks for each temperature benchmark. In a further development , starting at the municipalities level, we perform a meta-analysis of this association, first by macro regions of Brazil and second for the whole country, from which we derive a functional form for the relative risk that each temperature percentile has in making a dengue-case a possible hospitalization for dengue
Dengue has been a disease present in the Brazilian ecological and epidemiological scenarios since the mid-twentieth century. From the 1990s, the cases that, before, were sporadic,and distributed without any clear association to population and territory variables, began to exhibit a new dynamic. In 2001, by a law that made notification of dengue cases mandatory, notifications of the disease exploded and dengue began to be seen as a public health problem. Dengue is a disease transmitted by the mosquito Aedes Aegypti. It has a clear seasonal component. However, even if it is understood that certain seasonality is derived from climate, we still lack better quantitative analyses of the effects of climate on dengue epidemics. In this thesis, time-series of dengue cases and essential climate variables (ECV), such as temperature and precipitation, are analysed using convergent cross-mapping (CCM). CCM is based on the result of the Takens theorem which states that the attractor of a dynamical system can be reconstructed from the time-series of one of the variables of the system, and the time series of this same variable with time delays. Based on this result, methods for determining causal relations have been developed. We use these methods and show that, in the case of the city of Rio de Janeiro, precipitation is the most important ECV relent for dengue epidemics.. Dengue, in a less common way, can lead to hospitalization according to previous infection history and weather conditions. Using time-series of hospitalized dengue cases and temperature series, we explore which are the risks of developing a more severe dengue condition given temperature exposure. To this end, we use distributed lag non-linear models (DLNM), which, by using delays of a series of an exposure factor in a generalized linear model with predictors of the response series, in our case the series of dengue hospitalizations, provides a statistical association between these factors. This allows us to account for relative risks for each temperature benchmark. In a further development , starting at the municipalities level, we perform a meta-analysis of this association, first by macro regions of Brazil and second for the whole country, from which we derive a functional form for the relative risk that each temperature percentile has in making a dengue-case a possible hospitalization for dengue
Descrição
Palavras-chave
Causalidade, Análise de séries temporais, Modelos estátiscos
Como citar
SILVA, Rafael Lopes Paixão da. Dengue dynamics under climate drivers: analysis with ecological and epidemiological frameworks. 2023. Tese (Doutorado em Física) - Instituto de Física Teórica, Universidade Estadual Paulista, São Paulo, 2022