我国流行性乙型脑炎时空分布特征及风险预测研究

发布时间：2018-04-12 11:34

本文选题：流行性乙型脑炎 + 时空聚集区　；参考：《中国人民解放军军事医学科学院》2014年硕士论文

【摘要】：背景：流行性乙型脑炎(Japanese encephalitis, JE),简称乙脑，由乙型脑炎病毒（Japanese encephalitis virus, JEV）引起的急性中枢神经系疾病，我国法定乙类传染病。乙脑是一种蚊媒传染病，通过被感染的蚊虫叮咬而传播，猪是其重要的扩增宿主，，迁徙的水鸟和蝙蝠等是主要的储存宿主，并能携带病毒进行远距离扩散。人对JEV普遍易感，感染后有较广的疾病谱：轻症可表现为无症状感染或是流感样症状；典型患者以高热、意识障碍、抽搐、脑膜刺激等神经症状为主要特征；重症患者可表现为呼吸衰竭甚至死亡。WHO最近研究显示，亚洲和西太平洋地区超过24个国家存在乙脑的流行，全球每年约有67,900病例发生，其中50%发生在中国。自20世纪70年代我国开始接种乙脑疫苗以来，乙脑发病率从1971年的20.92/10万下降到2008年的0.23/10万。目前，我国乙脑的控制已经取得了举世瞩目的成就，但也存在一些不容忽视的问题。比如，我国沿海地区乙脑发病已经得到有效控制，但云南、贵州、四川、广西等地乙脑的发病率还较高，与WHO提出的在2015年将小于15岁儿童乙脑发病率控制在0.5/10万以下的目标还有一定距离。其次，JEV流行区域已经扩大。曾经被认为是JEV空白区的西藏，已经从三带喙库蚊体内分离到基因I型病毒，在猪的血清标本中检测到IgM抗体和健康人的血清中检测到中和抗体。再者，近年来我国各地时有乙脑暴发，并呈现出一些新的特点：山西运城2006年7-8月乙脑暴发，病死率高达28.8%，成年型病例增多（86%的病例是30岁以上的成年人），并且同时分离到基因I型和III型，而这些现象在以往的暴发中是未出现的。尽管我国乙脑发病数大幅下降，但是由于20-30%病人死亡，30-50%幸存者有神经系统后遗症，给家庭和社会带来极大的精神和疾病负担，尤其值得注意的是乙脑是威胁云南儿童死亡的第二位传染病。综上可见，乙脑仍然是我国严重的公共卫生问题。近年来，空间信息技术得到迅猛发展，广泛应用于血吸虫病、疟疾、肾综合征出血热、蜱媒疾病和其他自然疫源性疾病的医学地图编制，探测时空分布规律、识别潜在的高风险地区和危险因素分析等的研究中。目前我国乙脑流行特征的研究主要是对局部地区长期，或全国范围内短期监测数据的分析，这就缺乏一个长期整体的认识。我国乙脑整体的时空分布格局、热点区域变迁问题、危险因素及风险预测等方面的研究非常少，存在较大的空白。近年来，我国乙脑风险预测的研究处于起步阶段，并且是从大尺度上进行研究，只能提供一个整体的趋势情况，精度欠缺，对于我国乙脑的靶向控制提供的参考非常有限。综上所述，有必要了解我国乙脑小尺度上的时空分布格局，找出发病的热点区域，分析高发区乙脑流行的主要影响因素，并预测我国乙脑的发生风险，为乙脑防控提供科学依据。目的：了解我国乙脑的流行趋势及变化特征，探讨我国乙脑时空分布模式及发病的热点区域。阐明不同高发区乙脑发生与环境因素的关系，识别其发生的主要驱动因素及其滞后作用。综合分析气象、地理和环境因素对我国乙脑发生的影响，建立高分辨率的乙脑发生风险预测模型，评估乙脑发生的空间差异，为我国乙脑预防控制提供科学依据。方法： 1.收集2002年1月-2010年12月全国各区县每月监测数据，运用描述性流行病学方法分析我国乙脑发生和死亡病例的流行特征；将区县发病数与相应的行政边界地图进行空间关联，绘制动态发病率地图，进行空间全局自相性分析（spatial autocorrelation），局部自相关性分析（Local Indicator of Spatial Association,LISA）和时空聚集性分析（spatial scan statistics）。以上过程由SAS9.2、SPSS16.0、GeoDaTM0.95i、ArcGIS9.3和SaTScan9.1.1软件实现。 2.收集高发地区（遵义市和毕节市）2002年1月-2011年12月乙脑每月监测数据，在调整乙脑发生的自相关性、季节性和长期趋势后，构建乙脑的发生和环境因素关系的零膨胀负二项回归模型(Zero-inflated negative binomial，ZINB)，并且与负二项回归模型（negative binomial, NB）、零膨胀Poisson模型（Zero-inflated Poisson, ZIP）进行比较。以上过程由Stata11软件实现。 3.收集具有详细家庭地址的乙脑病例个案信息。应用地理信息系统（Geographic information system, GIS）,将乙脑病例进行空间定位，并与我国气象因素（平均温度、最低温度、最高温度、相对湿度和降雨量）、海拔、土地利用、归一化植被指数、猪密度和人口密度1×1km分辨率的栅格图像一起构建基于最大熵原理（Maximum entropy, ME）的生态位模型（Ecological Niche Model,ENM），分析环境变量和乙脑发生之间的关系，评估各变量的贡献百分比，预测我国乙脑发生的风险。以上过程由MaxEnt3.3.3k和ArcGIS9.3软件实现。结果： 1.2002-2010年我国共报告乙脑病例48,892例，男女性别比为1.57:1；病例以15岁以下的儿童为主，占发病总数的87.40%，但近年来成年型乙脑（年龄大于等于40岁）的构成比例有上升的趋势（Cochran-Armitage trend test,Z=9.60，p0.001）。报告发病率以小年龄组较高，10岁的儿童每组年均发病率均大于1/10万。2005-2010年我国共报告乙脑死亡病例1,531例，男女性别比为1.29:1，女性病死率高于男性（5.83%Vs.4.79%, χ2=15.30,p0.001）。以10岁分组发现，乙脑的病死率有随年龄上升的趋势（Cochran-Armitage trend test, Z=13.06,p0.001），70岁以上的老人病死率高达14.67%。外籍人士病死率高于我国居民（12%Vs.5.77%，χ2=4.71,p=0.03）。根据乙脑流行特征（2002年为一个暴发期，随后发病数迅速下降，2006年又存在一个暴发期，病例达到峰值，随后迅速下降，2007-2010年乙脑发生保持在一个较低的水平上下波动），将2002-2010分为四个时间段：2002，2003-2005，2006，2007-2010进行自相关和聚集性分析，发现我国乙脑的发生存在显著的空间自相关性。LISA分析发现高-高聚集区主要集中在我国西南地区，同时有向中部扩大的趋势。选择最大时间窗口为10%的研究时间，最大的空间窗口为5%的人口数的椭圆形扫描窗口进行时空聚集性分析，发现一级聚集区在我国西南地区的7月和8月，覆盖的区县数目逐年增多。各个一级聚集区呈现较高的发病率水平（1.79/10万-3.27/10万），这些仅占全国3.88%-4.97%人口数的区县有全国同期30.17%-77.59%的乙脑病例。 2.在调整乙脑发生的自相关性、季节性和长期趋势后，位于一级聚集区内，以中低海拔为主的遵义市和以高海拔为主的毕节市进行ZINP多因素分析结果显示：遵义市乙脑发生与滞后3月的相对湿度和滞后4月的降雨呈负相关，其它因素保持不变时，滞后3月的相对湿度每增加1%，乙脑发生减少5.66%（95%CI：1.67%-9.45%）；其它因素保持不变时，滞后4月的降雨量每增加1mm，乙脑发生减少0.05%（95%CI:0.01%-0.08%）；毕节市乙脑发生与滞后1月最高温度呈正相关，其它因素保持不变时，滞后1月的最高温度每增加1oC，乙脑发生增加16.47%（95%CI：5.52%-28.55%）。建立的乙脑预测模型具有较好的拟合和预测效果，预测值的残差均为白噪声序列，残差的自相关和偏自相关图也显示残差为随机序列。 3.结合地理信息技术和遥感技术，应用基于最大熵原理的生态位模型分析我国乙脑发生与环境因素的关系，并对下一年乙脑发生进行预测，结果显示：最低温度贡献最大（17.94%-38.37%），人口密度、平均温度和海拔次之，分别是15.47%-21.82%，3.86%-21.22%，12.05%-16.02%；月平均温度12oC或最低温度高于-8oC的地区有乙脑的发生，月平均降雨量在80-120mm间或NDVI值达到150的地区，乙脑发生风险较高；相对湿度大于65%后，随着它的增加发生风险也逐渐增加，85%时达到峰值，随后保持平稳或小幅度下降；海拔与乙脑发生之间呈现负相关，高海拔地区乙脑发生的风险较低；土地利用中的灌溉农地、雨浇地和建筑用地与乙脑发生之间有较强的相关性；乙脑发生风险随着猪密度和人口密度的增加而增加，当分别达到400头/平方公里和2500人/平方公里时乙脑发生风险保持稳定。每个模型的受试者工作特征曲线下面积（the area under the curve,AUC）在0.82-0.91的范围，假阴性率在5.44-7.42%的范围，说明模型预测效果好。模型预测我国乙脑发生的高风险区主要在我国西南地区和中部地区。利用乙脑发病数和模型预测风险进行叠加分析，发现病例的发病地点与预测的风险高低一致性很高，病例基本处于中、高风险区域内或附近，大多数的病例（接近60%）发生在高风险地区，而这些地区仅占了不超过6%的国土面积。结论：本研究描述了我国乙脑流行规律和分布特征。利用GIS强大的数据分析与可视化功能，同时考虑疾病发生时间和空间属性的时空扫描技术，首次从区县尺度上分析我国乙脑时空分布格局，确定了我国乙脑发生具有空间自相关性和高风险的一级聚集区和二级聚集区。构建ZINB，分析高发区乙脑的发生与环境因素的关系，定量分析不同地区的主要影响因素和滞后作用，并进行了短期的时间尺度的预测。首次构建全国范围内高分辨率（1×1km）的空间风险预测模型，探讨了影响我国乙脑发生的主要环境因素，发现乙脑发生的高风险地区和潜在风险地区。研究结论为我国乙脑的监测、蚊媒控制、强化免疫接种等提供科学依据。
[Abstract]:Background :

Japanese encephalitis ( JE ) , known as the Japanese encephalitis virus ( JE ) , is an acute central nervous system disease caused by Japanese encephalitis virus ( JEV ) . The encephalitis B is a mosquito - borne infectious disease .
Typical patients were characterized by high fever , conscious disturbance , convulsions , meninges stimulation and other neurological symptoms .
Severe cases may be manifested as respiratory failure or even death . Recent studies by the WHO suggest that more than 24 countries in Asia and the Pacific have a prevalence of encephalitis B , around 67,900 worldwide each year , of which 50 per cent occurred in China .

The incidence of encephalitis B in China has been reduced from 20.92 per 100,000 in 1971 to 0.23 per 100,000 in 2008 .

In recent years , the development of spatial information technology has been widely used in the study of schistosomiasis , malaria , hemorrhagic fever with renal syndrome , tick - borne diseases and other natural epidemic diseases .

Purpose :

To understand the epidemic trend and the changing characteristics of the encephalitis B in China , the paper discusses the relationship between the occurrence of encephalitis B and the environmental factors in China , and identifies the main driving factors and its hysteresis effect . The effects of meteorological , geographical and environmental factors on the development of the encephalitis B in China are analyzed . The spatial difference of the development of the encephalitis B is assessed and the scientific basis for the prevention and control of the encephalitis B in China is provided .

Method :

1 . Collect the monthly monitoring data from January 2002 to December 2010 , and use descriptive epidemiological method to analyze the epidemic characteristics of the occurrence and death of encephalitis B in China ;
spatial correlation of the number of occurrence of district and county with corresponding administrative boundary map is carried out to map dynamic morbidity map , spatial global self - correlation analysis ( spatial correlation ) , local self - correlation analysis ( LISA ) and spatial scan statistics are carried out . The above procedure is implemented by SAS9.2 , SPSS 16.0 , GeoDaTM0.95i , ArcGIS 9.3 and SaTScan9.1 . 1 software .

2 . Collect the monthly monitoring data from Jan . 2002 to December 2011 in the high incidence area . After adjusting the auto - correlation , seasonal and long - term trend of the development of the encephalitis B , the zero - expansion negative binomial regression model ( ZINB ) is constructed and compared with the negative binomial regression model ( NB ) and the zero - expansion Poisson model ( Zero - Expansion Poisson , ZIP ) . The above process is realized by the Stata11 software .

3 . To collect the case information of Japanese encephalitis cases with detailed family address . Using Geographic Information System ( GIS ) , the case of encephalitis B is spatially located , and the ecological niche model ( ENM ) based on Maximum Entropy ( ME ) is constructed along with the meteorological factors ( average temperature , minimum temperature , maximum temperature , relative humidity and rainfall ) , altitude , land use , normalized vegetation index , pig density and population density of 1 脳 1 km . The contribution percentage of each variable is assessed and the risk of the development of Japanese encephalitis is predicted .

Results :

1 . In 2002 - 2010 , 48,892 cases of encephalitis B were reported , and the sex ratio of male and female was 1 . 57 : 1 ;
Cases were dominated by children under the age of 15 , accounting for 87.40 % of the total morbidity , but in recent years , the proportion of adult type B ( > = 40 years ) had a rising trend ( ran ran - Armitage trend test , Z = 9.60 , p0.001 ) . In 2005 - 2010 , 1 , 531 male and female sex ratio was 1 . 29 : 1 , and the mortality of female was higher than that of men ( 5.83 % vs . 4.79 % , 蠂 2 = 15.30 , p0.001 ) . A 10 - year - old group found that the mortality rate of the encephalitis B increased with the age ( ran ran - Armitage trend test , Z = 13.06 , p0.001 ) . The mortality of the elderly over 70 years of age was 14.67 % . The mortality of foreigners was higher than that in our country ( 12 % vs . 5.77 % , 蠂 2 = 4.71 , p = 0.03 ) . In 2002 , 2003 - 2005 , 2006 and 2007 - 2010 , the number of districts and counties covered by the LISA showed a high incidence level ( 1.79 / 100000 - 3.27 / 100000 ) , which accounted for 30.17 % - 77.59 % of the population in the same period .

2 . After adjusting the auto - correlation , seasonal and long - term trend of the development of the encephalitis B , the ZINP multi - factor analysis results show that the relative humidity and the lag of the middle and low altitudes are negative correlated with the rainfall in April and the other factors remain unchanged . When the other factors remain unchanged , the relative humidity after the lag is increased by 1 % , the incidence of encephalitis B is 5.66 % ( 95 % CI : 1.67 % - 9.45 % ) ;
When other factors remained unchanged , the rainfall in April was increased by 1mm and the incidence of encephalitis B decreased by 0.05 % ( 95 % CI : 0.01 % - 0.08 % ) ;
In Bijie City , the highest temperature was positively correlated with the highest temperature in the first month , while the other factors remained unchanged , the highest temperature in January was increased by 16.47 % ( 95 % CI : 5.52 % - 28.55 % ) . The established model has better fit and prediction effect , the residual of prediction value is white noise sequence , the self - correlation and partial autocorrelation of residual are also the random sequence .

3 . Combined with the geographic information technology and remote sensing technology , the relationship between the development of Japanese encephalitis and environmental factors was analyzed by using niche model based on the principle of maximum entropy . The results showed that the minimum temperature contribution ( 17.94 % - 38.37 % ) , population density , average temperature and altitude were 15.47 % - 21.82 % , 3.86 % - 21.22 % , 12.05 % - 16.02 % , respectively .
The average monthly temperature is 12oC or the lowest temperature is higher than -8oC . The average monthly rainfall is between 80 - 120 mm or 150 , and the risk of encephalitis B is higher .
When the relative humidity is greater than 65 % , the risk is increased gradually with the increase of the relative humidity , reaching the peak at 85 % , followed by a smooth or small decrease ;
There was a negative correlation between altitude and the incidence of encephalitis B , and the risk of encephalitis B in high altitude area was lower .
There was a strong correlation between irrigated farmland , rain - irrigated land and building land and encephalitis in land use .
The risk of encephalitis B increases with the increase of the density of pigs and population density . The risk of encephalitis B is stable when 400 heads / km2 and 2500 people per square kilometer are respectively reached . In each model , the area under the curve ( AUC ) ranges from 0.82 to 0.91 , and the false negative rate is in the range of 5.44 - 7.42 % .

Conclusion :

This paper describes the epidemic regularity and distribution of Japanese encephalitis B . Based on the data analysis and visualization functions of GIS , the spatial and temporal distribution pattern of the disease occurrence time and space attribute is analyzed . The relationship between the occurrence of the encephalitis B and the environmental factors is analyzed . The main influencing factors and lagging effects in different regions are analyzed quantitatively .

【学位授予单位】：中国人民解放军军事医学科学院
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R512.32

【参考文献】