Study On Epidemic Characteristics And Prediction Model Of Influenza In Hohliot From 2014 To 2017

Objective To analyze the etiological monitoring results and epidemiological characteristics of influenza-like illness（ILI）in Hohhot from 2014 to 2017,to understand the molecular evolution characteristics of HA and NA of influenza virus in Hohhot from 2015 to 2017,to explore the correlation between influenza-like illness and meteorological and air quality factors,and to construct a prediction model of influenza-like illness by time series analysis,so as to provide scientific basis for influenza prevention and control work and early warning prediction.Methods The surveillance datas of influenza-like cases in Hohhot from 2014 to 2017 were collected,descriptive statistical analysis of epidemiology was conducted on ILI monitoring data.Chi-square test was used for the rate comparison.One H3N2 representative strain from 2015 to 2017 every year was selected for HA and NA gene sequencing,spliced and sorted the sequencing products by using Seqman in DNAStar software to obtain HA and NA gene full-length sequences,the HA and NA sequences of the reference strain and vaccine strain were downloaded from the gene sequence database（Gen Bank）of the National Biological Information Center and the global shared bird flu database（GISAID）.The Muscle method in MEGA 7.0 software was used for sequence alignment,the Neighor-Joining（NJ）method was used to construct the phylogenetic tree of HA and NA genes,and Bootstrap method was used for testing with 1000repetitions.The meteorological data and air quality factors data from 2014 to 2017were collected to analyze the correlation between meteorological factors and air quality factors with influenza-like cases.The seasonal autoregressive moving average model for influenza-like cases in the region was established.The goodness of fit and prediction accuracy of the models before and after the introduction of external factors were analyzed and compared.Results 1.The total number of ILI cases in Hohhot from 2014 to 2017 was 78,093.The average reported percentage of ILI（ILI%）in four years was 2.13%.The report rate of ILI（%）declines year by year in four years andthe difference in reported ILI（%）in four years was statistically significant（?²=715.17,P=0.000）.2.Influenza-like cases in Hohhot from 2014 to 2017 showed a high incidence in winter and spring and a stable state in summer and autumn.The number of children aged 0～4 years was the largest,with 53,727 cases in all,accounting for68.8%of the total influenza-like cases,while the 15～24 years old group reported the least with 1525 cases,accounting for 1.95%of the total influenza-like cases.3.A total of 4,178 influenza-like cases were isolated in Hohhot from 2014 to 2017,with451 isolated positive cases,the average isolation positive rate was 10.79.Influenza A and B viruses were mixed and prevalent in four years,but the dominant strains prevalent each year were different:the H3 subtype and B Yamagata subtype were mixed dominant strains in 2014,the H3 subtype in 2015,B Victoria subtype in 2016and new H1 subtype in 2017.4.Influenza virus was isolated in different age groups.The highest isolation positive rate was 18.25%in the 15-24 years old group,while the lowest isolation positive rate was 6.59%in the over 60 years old group.The difference of virus isolation positive rate was statistically significant（?2=57.19,P=0.000）.Comparison of virus isolation between different genders:male’s virus isolation positive rate was 10.65%,female’s virus isolation positive rate was 10.98%,the difference had no statistically significant（?2=0.07,P=0.79）.5.Phylogenetic tree analysis of HA and NA genes of H3N2 influenza virus strains from 2015 to 2017in Hohhot showed that the isolates from different monitoring years were distributed in different branches and all isolates were not on the same branch as the recommended vaccine strains from the corresponding years,suggesting that HA and NA genes of influenza virus strains in this study had some variation,and the influenza vaccines from the corresponding years had poor protection against H3N2 influenza virus.6.Correlation analysis of ILI with meteorological factors and air quality factors showed that only temperature and humidity were related to ILI.The datas were fitted with different orders of single variable ARIMA（p,d,q）（P,D,Q）_s,of which single variable ARIMA（1,0,1）（0,1,1）₅₂ model had the highest stable R²value0.733,the lowest RMSE value63.031.The average weekly temperature lag of 1 week was included in the SARIMA（1,0,1）（0,1,1）₅₂prediction model.The stable R² and RMSE of the model after the weekly average of temperature lag by one week were 0.745 and 61.967respectively,indicating that the goodness of fit and prediction accuracy of the model were improved after the introduction of meteorological parameter.Conclusion 1.Influenza-like cases in Hohhot from 2014 to 2017 were generally high in winter and spring and stable in summer and autumn.Most influenza-like cases were among children aged 0～4 years.Monitoring and prevention of this population should be strengthened.2.Hohhot presented a mixed epidemic of influenza A and B viruses from 2014 to 2017.The dominant strains prevalent each year were different,showing the characteristics of alternating occurrence of one with the other.3.The phylogenetic tree of influenza virus HA and NA in this study showed that the corresponding annual influenza vaccine had relatively low protective effect on H3N2influenza virus in Hohhot.4.Partial correlation analysis between influenza-like cases and air quality factors had no statistical significance.Weekly mean temperature and weekly mean humidity were related to the epidemic of influenza-like cases.The SARIMA model included weekly mean temperature could be used as a technical method for short-term prediction of influenza epidemic.