RESUMO
BACKGROUND: To our knowledge, no study has assessed the association between heatwaves and risk of hospitalization and how it may change over time in Brazil. We quantified the heatwave-hospitalization association in Brazil during 2000-2015. METHODS AND FINDINGS: Daily data on hospitalization and temperature were collected from 1,814 cities (>78% of the national population) in the hottest five consecutive months during 2000-2015. Twelve types of heatwaves were defined with daily mean temperatures of ≥90th, 92.5th, 95th, or 97.5th percentiles of year-round temperature and durations of ≥2, 3, or 4 consecutive days. The city-specific association was estimated using a quasi-Poisson regression with constrained distributed lag model and then pooled at the national level using random-effect meta-analysis. Stratified analyses were performed by five regions, sex, 10 age groups, and nine cause categories. The temporal change in the heatwave-hospitalization association was assessed using a time-varying constrained distributed lag model. Of the 58,400,682 hospitalizations (59% women), 24%, 34%, 21%, and 19% of cases were aged <20, 20-39, 40-59, and ≥60 years, respectively. The city-specific year-round daily mean temperatures were 23.5 ± 2.8 °C on average, varying from 26.8 ± 1.8 °C for the 90th percentile to 28.0 ± 1.6 °C for the 97.5th percentile. We observed that the risk of hospitalization was most pronounced for heatwaves characterized by high daily temperatures and long durations across Brazil, except for the minimal association in the north (the hottest region). After controlling for temperature, the association remained for severe heatwaves in the south and southeast (cold regions). Children 0-9 years, the elderly ≥70 years, and admissions for perinatal conditions were most strongly associated with heatwaves. Over the study period, the strength of the heatwave-hospitalization association declined substantially in the south, while an apparent increase was observed in the southeast. The main limitations of this study included the lack of data on individual temperature exposure and measured air pollution. CONCLUSIONS: There are geographic, demographic, cause-specific, and temporal variations in the heatwave-hospitalization associations across the Brazilian population. Considering the projected increase in frequency, duration, and intensity of heatwaves, future strategies should be developed, such as building early warning systems, to reduce the health risk associated with heatwaves in Brazil.
Assuntos
Poluição do Ar/efeitos adversos , Exposição Ambiental/efeitos adversos , Hospitalização/tendências , Temperatura Alta/efeitos adversos , Análise de Séries Temporais Interrompida/tendências , Adolescente , Adulto , Idoso , Brasil/epidemiologia , Cidades/epidemiologia , Feminino , Humanos , Masculino , Pessoa de Meia-Idade , Fatores de Risco , Fatores de Tempo , Adulto JovemRESUMO
Introduction: Online cigarette dealers have lower prices than brick-and-mortar retailers and advertise tax-free status.1-8 Previous studies show smokers search out these online alternatives at the time of a cigarette tax increase.9,10 However, these studies rely upon researchers' decision to consider a specific date and preclude the possibility that researchers focus on the wrong date. The purpose of this study is to introduce an unbiased methodology to the field of observing search patterns and to use this methodology to determine whether smokers search Google for "cheap cigarettes" at cigarette tax increases and, if so, whether the increased level of searches persists. Methods: Publicly available data from Google Trends is used to observe standardized search volumes for the term, "cheap cigarettes". Seasonal Hybrid Extreme Studentized Deviate and E-Divisive with Means tests were performed to observe spikes and mean level shifts in search volume. Results: Of the twelve cigarette tax increases studied, ten showed spikes in searches for "cheap cigarettes" within two weeks of the tax increase. However, the mean level shifts did not occur for any cigarette tax increase. Conclusion: Searches for "cheap cigarettes" spike around the time of a cigarette tax increase, but the mean level of searches does not shift in response to a tax increase. The SHESD and EDM tests are unbiased methodologies that can be used to identify spikes and mean level shifts in time series data without an a priori date to be studied. SHESD and EDM affirm spikes in interest are related to tax increases. Implications: ⢠Applies improved statistical techniques (SHESD and EDM) to Google search data related to cigarettes, reducing bias and increasing power ⢠Contributes to the body of evidence that state and federal tax increases are associated with spikes in searches for cheap cigarettes and may be good dates for increased online health messaging related to tobacco.
Assuntos
Algoritmos , Comércio/tendências , Internet/tendências , Análise de Séries Temporais Interrompida/tendências , Impostos/tendências , Produtos do Tabaco , Comércio/economia , Análise de Dados , Humanos , Internet/economia , Análise de Séries Temporais Interrompida/economia , Impostos/economia , Produtos do Tabaco/economiaRESUMO
BACKGROUND: The effect of socioeconomic stressors on the incidence of cardiovascular disease (CVD) is currently open to debate. Using time-series analysis, our study aimed to evaluate the relationship between unemployment rate and hospital admission for acute myocardial infarction (AMI) and stroke in Brazil over a recent 11-year span. METHODS AND RESULTS: Data on monthly hospital admissions for AMI and stroke from March 2002 to December 2013 were extracted from the Brazilian Public Health System Database. The monthly unemployment rate was obtained from the Brazilian Institute for Applied Economic Research, during the same period. The autoregressive integrated moving average (ARIMA) model was used to test the association of temporal series. Statistical significance was set at p<0.05. From March 2002 to December 2013, 778,263 admissions for AMI and 1,581,675 for stroke were recorded. During this time period, the unemployment rate decreased from 12.9% in 2002 to 4.3% in 2013, while admissions due to AMI and stroke increased. However, the adjusted ARIMA model showed a positive association between the unemployment rate and admissions for AMI but not for stroke (estimate coefficient=2.81±0.93; p=0.003 and estimate coefficient=2.40±4.34; p=0.58, respectively). CONCLUSIONS: From 2002 to 2013, hospital admissions for AMI and stroke increased, whereas the unemployment rate decreased. However, the adjusted ARIMA model showed a positive association between unemployment rate and admissions due to AMI but not for stroke. Further studies are warranted to validate our findings and to better explore the mechanisms by which socioeconomic stressors, such as unemployment, might impact on the incidence of CVD.
Assuntos
Análise de Séries Temporais Interrompida/tendências , Infarto do Miocárdio/epidemiologia , Admissão do Paciente/tendências , Acidente Vascular Cerebral/epidemiologia , Desemprego/tendências , Adulto , Idoso , Brasil/epidemiologia , Feminino , Hospitalização/tendências , Humanos , Análise de Séries Temporais Interrompida/métodos , Masculino , Pessoa de Meia-Idade , Infarto do Miocárdio/diagnóstico , Infarto do Miocárdio/economia , Fatores Socioeconômicos , Acidente Vascular Cerebral/diagnóstico , Acidente Vascular Cerebral/economiaRESUMO
Two smoothing strategies combined with autoregressive integrated moving average (ARIMA) and autoregressive neural networks (ANNs) models to improve the forecasting of time series are presented. The strategy of forecasting is implemented using two stages. In the first stage the time series is smoothed using either, 3-point moving average smoothing, or singular value Decomposition of the Hankel matrix (HSVD). In the second stage, an ARIMA model and two ANNs for one-step-ahead time series forecasting are used. The coefficients of the first ANN are estimated through the particle swarm optimization (PSO) learning algorithm, while the coefficients of the second ANN are estimated with the resilient backpropagation (RPROP) learning algorithm. The proposed models are evaluated using a weekly time series of traffic accidents of Valparaíso, Chilean region, from 2003 to 2012. The best result is given by the combination HSVD-ARIMA, with a MAPE of 0:26%, followed by MA-ARIMA with a MAPE of 1:12%; the worst result is given by the MA-ANN based on PSO with a MAPE of 15:51%.