Impact Of Vector And Meteorological Factors On Dengue Epidemic In Guangzhou And Establishment Of Forecasting Model

Dengue fever (DF) is an acute infectious disease spread widely in tropical and subtropical regions worldwide and caused by dengue virus transmitted by Aedes aegypti and Aedes albopictus. One of the four types of dengue virus (serotype 1,2,3 and 4) may cause asymptomatic infection, dengue fever or dengue hemorrhagic fever and even death. Dengue has become a harmfulkind ofinsect-borneviral disease with themost widely distribution and the most high incidence. Dengueis a serious public health problem inmany countriesand regions and one of the 17 kinds ofneglected tropical diseases listed by World Health Organization (WHO). The incidence of dengue increased significantly in recent decades in the world. An estimated 50 to 100 million people live in over 100 endemic countries and areas affected dengue each year and about 2.5 billion people accounting for more than 40% of theworld population with high risk of dengue.Seven epidemics of dengue appeared in China during 1978-1990. Since 1991, dengue outbreak occurred mainly in Guangdong province with a larger epidemic in 1995 and peaks in 4-7 years. Dengue cases and epidemics increased significantly in recently years in Guangdong Province since 2000. A dengue outbreak occurred in Yunnan from August to October 2013 caused by imported cases. The scale of dengue epidemic reached a new peak of 46,864 cases in 2014 after a pandemic in 1986. With global warming, widespread vector, urbanization and population growth, dengue will continue to be a serious threat to China in the future.Guangzhou has always been the focus of dengue control due to its high incidence of dengue. There were 15 popular years and seven outbreaks with more than 1000 cases since 1978. Dengue became more active in Guangzhou in the mid of 1980s and with lower intensity in the 1990s. A pandemic of dengue occurred in Shawan town of Panyu district in 1995 reached the historic peak. Then, four larger epidemics appeared from 2002 and reached the peak in 2014. All of the four types of dengue virus had appeared in Guangzhou, but DENV-1 strains became more prevalent in local cases after 1990.Guangzhou, as the capital city, China’s "South Gate" and an important national transportation hub, has more frequent population movements. Guangzhou has a subtropical monsoon climate with the annual average temperature of 21-23 ℃ and the average relative humidity of 75%, which are suitable for local dengue epidemic caused by imported cases from the endemic areas.Climate as one of the three transmission elements has the most significant impact on dengue. Aedes albopictus is the main vector in dengue transmission in Guangzhou. The significant seasonal fluctuation in dengue was mainly impacted by climate factors. Climate can have a major impact both directly and indirectly on the number of vector and dengue virus. Some previous studies have confirmed that climatic conditions affect mosquito breeding, maturity, viral replication, density and life cycle.Dengue is influenced by meteorological factors, so the relationship between dengue and climate changes attracted many scholars extensive concern. Application of models to analyze this problem is the most common method. A plenty of previous studies using different methods (including mathematical models, cross-correlation analysis and time series analysis, etc.) has assessed the relationship between dengue and climate. Autoregressive Integrated Moving Average Model (ARIMA) could estimate the incidence of dengue in a specific and small region by introducing meteorological parameters. Regression model and zero-inflated model (ZIM), which were used commonly for count data in previous studies, could get better conclusions in analyzing the linear or log-linear relationship between dengue and meteorological factors.However, some studies indicated that the effects of different ranges of temperatures on dengue occurrence are distinct. We speculated that the relationships between dengue incidence and some meteorological factors may be log non-linear. There is a need to determine the mechanism on how meteorological factors influence the development of dengue epidemics.Based on the 2014 dengue pandemic in Guangzhou, to indentify epidemic factors and to explore effective methods of early warning and assessment of control measures is essential. In this study, we try to analyze the epidemic characteristics of dengue in Guangzhou,2010-2014. ARIMA models and two models for count data were used to identify the impact of vector factors and meteorological factors on dengue epidemic, as well as explore the most suitable model. We also evaluated the prevention and control measures in dengue epidemic in 2014 from five aspects. The results could provide a reference for the effective control strategies of dengue and help to build predictive models in early warning and control of dengue in the future.Chapter 1 Study on epidemiological characteristics of dengue fever in recent 5 years in GuangzhouThrough analyzing the epidemic characteristics and spread patterns of dengue in Guangzhou during 2010-2014 and compare the similarities and differences in outbreaks since 2000, we explored the rules of dengue epidemic and provide a reference for the prevention and control of dengue.MethodsDengue epidemiology and etiology monitoring data in Guangzhou during 2010-2014 were collected. We analyze the characteristics and trends of dengue in recent five years, trends and dynamics with comparison between the 2014 and 2002, 2006 2013 epidemic years by establishing database and using the descriptive epidemiological method.Results1. Dengue cases reported in Guangzhou were 52,926 and 14 deaths in 37 years from 1978 to 2014, during which there were 15 epidemic years, seven outbreaks with more than 1000 cases and a peak in 2014. Every year local outbreaks and sporadic cases were caused by imported cases. The peak of incidence emerged every 3-5 years overall.2. There were 1478 reported local dengue cases in Guangzhou during 2010-2013, with the incidence fluctuations of 0.31 per 100,000 to 9.85 per 100,000. The peak of incidence emerged from August to October and epidemic period involved 6-12 months with apparent seasonal fluctuations. Four serotypes of dengue were all prevalent in varying degrees, with DENV-1 more popular. The total reported cases during 2010-2014 in Guangzhou accounted for 79.86% (38818/48609) of Guangdong Province and 74.02% of China (38818/52445).3. The reported local cases were 37,340 in 2014, which was 2.40 times of the total cases in 1978-2013. The annual reported incidence rate was 290.11/100,000, while the mortality rate was 0.039/100,000. The 2014 epidemic lasted for 193 days could be divided into five periods. The sporadic period began from June 11 and lasted for five weeks with 43 reported cases. The rapid rising period lasted for nine weeks since July 13, accounting for 13.16% of the total reported cases. The peak of epidemic lasted for 4 weeks since September 14, accounting for 65.32% of the total reported cases. The rapid decline period lasted for six weeks since October 12, accounting for 21.10% of the total reported cases. The end of the period lasted for four weeks since November 23 and ended in December 20. All of the 12 districts were involved in dengue epidemic and the top five districts accounted for 82.02% of the total reported cases. The 2014 outbreak involved 159 Streets, accounting for 94.64% of the total streets.4. The 2014 epidemic had 5 different features compared with the 2002,2006, 2013 outbreaks. Firstly, the first local cases emerged early and 51 days ahead of 2013 outbreaks. Secondly, the total 2014 epidemic lasted for 193 days, while the peak only lasted for 40 days. Thirdly, the 2014 epidemic spread widely and with outbreaks of regional concentration. The epidemic started from Dongshan Street of Yuexiu district since June and extended to the whole city and countryside. Fourthly, the proportion of severe cases was higher than 2013 epidemic, while the mortality rate was much lower than other countries.ConclusionGuangzhou is the main area with dengue epidemic and the focus of prevention and control of dengue. Since 1978, the overall prevalence of dengue in Guangzhou was a moderate intensity with a peak occurring every 3-5 years. Every year local outbreaks and sporadic cases were caused by imported cases. In recent 5 years the reported cases in Guangzhou accounted for two-thirds of the total reported cases in China and Guangdong, respectively. The main serotype of the epidemic was DENV-l.The epidemic model had significant seasonal feature. The 2014 epidemic, as the highest incidence had some different features from previous years, such as the early emerging of first local cases, short peak of the epidemic and longer period of whole epidemic, the widely spread range with regional concentration of the outbreaks and high proportion of severe cases than in previous years, but low mortality.Chapter 2 Impact of mosquito density and meteorological factors on dengue fever in Guangzhou and Establishment of forecasting modelThrough establishing the ARIMA model, negative binomial regression model (NB) and zero-inflated negative regression model (ZINB), we analyzed mosquito density and meteorological factors on dengue epidemic in Guangzhou to provide reference and help to establish further early warning and predictive model.Methods1. We established database including of monthly cases of dengue, mosquito density surveillance data and meteorological variables during January 2006-November 2014 in Guangzhou.2. We used three models to analyze the impact of mosquito density and meteorological factors on dengue fever. Firstly, we used auto-correlation analysis and cross-correlation analysis to screen variables and the lag time to build and test the ARIMA model. Then, we used time series filtering method named Seasonal-Trend Decomposition Procedure (STL) to select variables significantly correlated with the dengue cases into the NB model. We selected minimum AIC model according to the Akaike information criterion (AIC). Finally, we introduced Breteau Index (BI) and adult mosquito density (ADI) as indicators of mosquito density, combining meteorological factors to build the ZINB models, respectively. We used Vuong test to test model and used Intraclass Correlation Coefficient (ICC) to verify the consistency of the actual and predicted cases.3. Software used in the study included SPSS 22.0, Stata 13.0, R 3.1.2 and Eviews8.0.Results1.39,697 local dengue cases were reported in Guangzhou during January 2006 to November 2014. Mosquito density had cyclical changes in the summer and early autumn, therefore BI values above 5 occurred from May to September. The average monthly rainfall and temperature variables had seasonal fluctuations in April-October, while interannual variability of the average monthly relative humidity and wind velocity were very little. ARIMA model analysis showed that current monthly BI had a positive impact on the cases of dengue (t=4.643, P<0.001). The average monthly maximum temperature (Tmax) was negatively correlated with the cases (t=-3.460, P =-3.460). ICC between actual incidence and predictive value was 0.905(95%CI: 0.864-0.934, P<0.001), which indicating good consistency.2. The current month’s BI, average temperature (Tave), previous month’s minimum temperature (Tmin), and Tave were positively associated with DF incidence. A threshold of 18.25℃ was found in the relationship between the current month’s Tmin and DF incidence. An increase of 1 in current month’s BI was associated with a 35.8% (95% confidence interval (CI),13.9%-62.0%),34.2% (95%CI,11.0%-62.3%), 28.0% (95%CI,8.1%-51.7%), and 30.1% (95%CI,10.4%-53.4%) increase of DF incidence based on Models 1,2,3, and 4, respectively. An increase of 1℃ in current month’s Tave and previous month’s Tmin and Tave was related to a 29.0%(95%CI, 13.9%-46.0%),51.8%(95%CI,30.9%-76.1%), and 52.8% (95%CI,32.5%-76.3%) increase of DF incidence, respectively, and an increase of 1℃ in current month’s Tmin was related to a 64.9% (95%CI,34.8%-101.7%) increase of DF incidence when current month’s Tmin was smaller than 18.25℃.3. The mean of monthly dengue cases was 371, with a standard deviation of 2337.29 and variance of 5,462,904, which was much larger than the mean.O statistic was 107190.90(P<0.001) indicated an overdispersion of dengue cases. The zero count accounted for 53.27% (57/107) of the total cases. When only selecting adult mosquito density index (ADI) into the ZINB model, we found that one month lagged ADI (lagADI) had a significant impact on monthly dengue incidence (Case).An increase of 1 in ADI was associated with a 28.3% (95%CI:13.9%-62.0%) increase of dengue cases. Wind had a significant impact on the Case (IRR=17.878 (95% CI:1.094-292.200). When selecting BI into the ZINB model, we found that season, BI, Wind and one month lagged BI had a positive impact on Case, while Tmin, one month lagged Hum and Wind had negative effect. An increase of 1 in BI was associated with a 79.0% (IRR=1.790,95%CI:1.382-2.318) increase of dengue cases. An increase of 1 in one month lagged BI was associated with a 70.5% (IRR=1.705, 95%CI:1.332-2.183) increase of dengue cases.ConclusionThree models could well explain the impact of mosquito density and meteorological factors on dengue cases.Time series ARIMA model analysis showed that BI and Tmax had positive and negative correlation with dengue cases, respectively. NB model showed that the current month’s BI, average temperature (Tave), previous month’s minimum temperature (Tmin), and Tave were positively associated with DF incidence. A threshold of 18.25℃ was found in the relationship between the current month’s Tmin and DF incidence. This helps us to better understand the impact of Tmin on dengue cases. ZINB model also identified ADI and BI could be used asindicators to predict and assessof dengue epidemic. Mosquito density and meteorological factors play a critical role in dengue transmission in Guangzhou. Our findings could be useful to develop a dengue early warning system and assist in taking effective control and prevention strategy in dengue epidemic.Chapter 3 Assessment of prevention and control measures on dengue fever in 2014 in GuangzhouWe aimed to evaluate the control measures and effect of 2014 dengue epidemic in Guangzhou to provide the basis forimproving dengue control strategies and promoting the control efficiency to save social resources.MethodsWe used descriptive epidemiology methods to analyze data of dengue epidemic, vector surveillance and control measuresin 2014. We evaluated the control measures and effect from 4 aspects:the trends of case and mosquito density, assessmentof the 4 unified anti-mosquito campaigns, quality control of dengue laboratory test items and comparation of the prevalence of dengue in other areas.Results1. The mosquito density during March to September 2014 was higher than the same period of 2010-2013in Guangzhou. Dengue cases showed a reverse trend with the eligible rate ofmosquito density. The eligible rate of mosquito density, BI, ADI and SSI rised from 33.33%,61.40%,36.11% and 50.00%on September 1 to 70%, 95.17%,81.66% and 70.00% on November 9, respectively. The effect of mosquito control was remarkable. BI could be used as early warning indicators in dengue epidemic.2. The average ADI of the 55 surveillance sites declined from 19.86/person·hour on September 28 to 1.73/person·hour on October 7 afteranti-mosquito action.The total eligible rate of mosquito density increased from 43.64% (24/55) to 78.18% (43/55). The Mosquito density of seven parks also declined substantially.3. During October 14-15, we collected 256 blood samples of dengue cases from13 hospitalsby using the same method for laboratory review. We found the compliance rate of dengue non-structural protein 1 (NS1) antigen by gold standard method was 100% and tested by ELISA was 94.55% with false positive rate of 6.98% and false negative rate of 4.08%. The compliance rate of dengue antibody positive rate was 84.06%, whilethe false positive rate reached to 31.43%. The compliance rate ofnucleic acid test was 78.72% and the false positive rate was 35.71%. Inconsistent results mainly existed in the nucleic acid and colloidal gold antibodydetection methods.We collected314 blood specimens, including 255 from Hospital inspection and 59 from healthy people. The false positive rate of NS1 antigen testwas 9.70% and the false negative rate was 10.70%. The consistency coefficientof NS1 antigen test and PCR. or antibody test was 0.80, suggesting that NS1 antigen test with higher value for dengue diagnosis.4. We found a big difference when we compared dengue epidemic curve of Guangzhou with dengue endemic countries, such as Malaysia, Singapore, Philippines, Laos and Brazil, which indicated that Guangzhou was not endemic. We Compared dengue epidemic and control measures in Guangzhou with Kaohsiung, which hadthe same latitude and longitude. The epidemic in Kaohsiungstarted from January with sporadic cases and increased doubling from May, then reached its peak in October with the incidence of 221.50 per 100,000, which was 1.60 times of Guangzhou (138.45 per 100,000) at the same period. The incidence of November in Kaohsiung was 198.15 per 100,000 and 24.3 times of Guangzhou (8.15 per 100,000). The epidemin in Guangzhou stopped on December 20, while Kaohsiung still with many reported cases lasted for two months and the peakduration 2 times of Guangzhou. The different characteristics of two modes related to the prevention and control measures taken by the two cities.ConclusionWe assessed the effect of 2014 dengue control measures and found thatdengue cases showed a reverse trend with the eligible rate ofmosquito density.The effect of mosquito control was remarkable. NS1 test can be used for dengue pathogen detection for its high diagnostic value with low false negative rate and false positive rate. The sensitivity of NS1 antigen test was higher than nucleic acid and antibody test, which indicating its usfulness for early case investigation. The different characteristics between Guangzhou and Kaohsiung related to the extraordinaryprevention and control measures taken by Guangzhou. Powerful intervention of social factors had played an important role in the prevention and control of dengue fever.