| Background:Influenza is an acute respiratory infectious disease caused by influenza virus that threatens human health all over the world.The high incidence of influenza in China is expanding from the eastern provinces to the central and western regions.In recent years,the incidence of influenza in Shaanxi Province has shown an upward trend.Especially in 2019,the number of influenza cases in Xi’an was 14 times that in 2018,and the incidence was 5.5 times that of the national influenza incidence in the same year.In December 2019,an outbreak of novel Coronavirus Disease 2019（COVID-19）caused nearly 7 million deaths worldwide.Xi’an,the capital of Shaanxi province,which borders Hubei,was one of the first cities to be hit by the epidemic.In order to control the spread of COVID-19 after the outbreak,Xi’an COVID-19 Prevention and Control Headquarters and Xi’an Center for Disease Control and Prevention have taken a series of non-pharmaceutical interventions（NPIs）against COVID-19 and achieved good results.These interventions largely contributed to the rapid decline of 98%in the number of influenza cases in Xi’an in 2020,and the low epidemic status remained until the end of2021.However,no studies have been conducted to quantitatively assess the specific effects of these interventions on influenza.It is of great importance to grasp the effects of COVID-19 prevention and control measures on the spread of influenza in Xi’an for the epidemic level and long-term development trend in the middle and later stages of the epidemic,as well as the research and deployment of relevant prevention and control work.Objectives:The purpose of this study was to systematically analyze the data of influenza cases and etiological detection in Xi’an from 2011 to 2021,and to explore the epidemic characteristics and changing trend before,during and after the COVID-19.On the basis of influenza epidemic trend,combined with local climate and meteorological data,influenza-related Internet data and other multi-source data,machine learning algorithm was used to construct a Bayesian Structural Time Series（BSTS）model and Interrupted Time Series Analysis（ITSA）,to predict the influenza epidemic of Xi’an in2020-2021 without the interference of the COVID-19 and corresponding NPIs.The study aimed to quantify the short-term and long-term effects of the NPIs against COVID-19 on influenza in Xi’an,providing data support and theoretical basis for the prevention of influenza virus and other respiratory infectious pathogens in the future.Methods:1.The surveillance data of influenza cases in Xi’an from 2011 to 2021 were obtained from Xi’an Center for Disease Control and Prevention;Descriptive epidemiological methods were used to analyze the changing trend of the epidemic characteristics of the three stages and compare the epidemic differences in the three stages.2.We collected and collated the climate and meteorological data,influenza-related Internet data and other multi-source data,combining with influenza case monitoring data of Xi’an,to construct influenza prediction model using BSTS and machine learning algorithm to predict the trend of influenza in 2020-2021;Three breakpoints were set based on the intensity of NPIs implemented in different phases of the COVID-19 outbreak,and ITSA was used to analyze the impact of NPIs against COVID-19 in different periods after the outbreak of COVID-19 on the changing trend of influenza epidemic;finally,we predicted the counterfactual epidemic curve of influenza cases in Xi’an from 2020 to 2021without the influence of COVID-19 and corresponding NPIs,and quantitatively evaluated the long-term effects of post-pandemic interventions.Results:1.Epidemiological characteristics of reported influenza cases in Xi’an from2011 to 2021From March 20,2011 to December 31,2021,a total of 197,535 influenza cases were reported in Xi’an.A total of 30,908 laboratory samples were collected,and 4,987 of them tested positive.The incidence of influenza in Xi’an increased exponentially from 2013 to2019,but after the implementation of the NPIs against COVID-19,the incidence of influenza in Xi’an decreased by 97.68%from 2020 to 2021.The peak of influenza epidemic was concentrated in winter and spring,and the top three affected counties were Yanta District,Chang’an District and Weiyang District.There were significant differences in the distribution of influenza cases in different age groups,genders and occupations in Xi’an,and the differences were statistically significant（P<0.001）.The number of male cases（104,486）was slightly higher than that of female cases（93,049）,with a male to female ratio of 1.12:1.The cases were mainly concentrated in the age group of 6-15 years（72,409 cases,36.66%）.The top five occupations were students,childcare children,diaspora children,housework and unemployment,and farmers.The results of laboratory analysis showed that a total of 30 904 samples of influenza-like illness were collected and tested,among which 4 983 samples tested positive for nucleic acid,with a comprehensive positive rate of 16.12%（4 983/30 904）.The influenza strains in Xi’an showed an alternate epidemic trend.In 2018,there was a mixed epidemic dominated by A H1 strain,and in2019,A H3 strain dominated.From 2020 to 2021,Victoria type B was predominant.2.Prediction of influenza epidemic trend after COVID-19 outbreakThe BSTS models were constructed with different variables,and the results showed that compared with only meteorological factors,the case data,meteorological factors and Internet-related data were included in the model at the same time,which was more fitting.Based on the data of influenza cases in Xi’an from 2011 to 2019,we predicted the influenza epidemic trend in the absence of non-drug interventions by using the constructed BSTS model:Aassuming no COVID-19 epidemic and no prevention and control measures were taken,the 2020 influenza epidemic period in Xi’an would be extended to the 7thto10thweek,and there would be a pandemic peak in the winter and spring of 2020-2021.In fact,the number of 2020 influenza cases dropped to zero by week 6,and the winter and spring 2020-2021 influenza peak did not occur.ITSA results showed that the inclusion probability of the slope change in the first phase was 0.54,while the inclusion probability of the immediate effect was 0.52,suggesting that the first phase had the greatest impact on the influenza epidemic in the short term and in the long term.The long-term impact of COVID-19 interventions on influenza in Xi’an was determined by comparing the cumulative effect between the predicted value and the real value using the data of influenza incidence in Xi’an.The results showed that the difference between the actual and predicted influenza incidence in Xi’an from 2020 to 2021 was2,101/100,000,which calculated that the number of influenza cases in Xi’an decreased by about 270,000.Conclusions:1.During 2011-2021,the reported influenza cases in Xi’an showed differences in time,space and epidemic characteristics among populations.In 2019,the reported incidence of influenza outbreak in Xi’an was far higher than the seasonal epidemic level,reaching nearly 100 times that of previous years,and the epidemic time was longer than the epidemic period in previous years.The dominant strains of each subtype of influenza virus showed an alternating trend of A virus and B virus,among which influenza A virus was the dominant strain in Xi’an.2.A BSTS prediction model based on multi-source big data such as influenza case data,climate and meteorological data and influenza-related Internet data was successfully constructed has better prediction effect than the traditional model that only includes case data and climate and meteorological data..3.The results of ITSA analysis showed that non-drug interventions had significant long-term and short-term control effects on the influenza epidemic in Xi’an,which could rapidly reduce the incidence of influenza in the short term,and long-term intermittent interventions could control the epidemic of influenza in the population.The influence of non-drug interventions on the epidemic of COVID-19 was heterogeneous in both population and spatial distribution. |
PDF Full Text Request