【作者】 龙璐；

【导师】 聂绍发；

【作者基本信息】 华中科技大学 ， 流行病与卫生统计学， 2013， 博士

【摘要】 研究目的通过对1990年-2010年海南省疟疾疫情时空分析及影响因素的研究，为该省及类似地区疟疾防控提供科学依据。通过尝试时间空间分析技术与统计分析方法在疟疾研究领域的综合应用，为类似研究提供方法学参考。研究方法收集整理1990-2010年海南省疟疾疫情监测数据，将全国1:100万电子地图加工处理为1:100万海南省市县边界图。提取海南省各县市的气温、降水量、湿度等气象与环境数据，人口、GDP、卫生、人民生活指标等数据，建立海南省市县综合信息数据库。应用时空扫描统计量、时间序列分析方法、主成分分析以及多元线性回归分析方法对资料进行分析。所采用的软件包括Excel2003、SatScan9.0、SPSS13.0、Mapinfo7.0等统计软件。研究结果1.1990年-2010年海南省疟疾呈整体下降的趋势，总体来说，海南省疟疾发病最高峰为7、8月份，2月份为全年发病最低峰。海南省西南部地区是疟疾疾病的高发区。人群分布，30-50岁男性为高发人群，农民、工人及民工占总报告人数的85.11%。2.运用时空重排扫描统计量对2005-2010年疟疾发病数据进行分析发现，2005-2010年海南省可能存在7个疟疾聚集区域（P<0.05）。对2010年疟疾发病数据进行时空扫描发现有3个聚集区域（P<0.05）。运用2005年1月-2009年12月建立的ARIMA模型对海南省2010年1月份的疾病发病率进行预测，预测值为0.15/10万，95%的可信区间为[-1.04,1.33]，实际检测发病率为0.17/10万，疟疾月发病率实际值落入了预测值的95%可信区间内，预测相对误差为11.8%。将2010年1月份疟疾实际发病率纳入时间序列模型，重新拟合，并对2010年2-12月份疟疾发病率进行预测，实际疟疾发病率比理论疟疾发病率平均下降了85.75%。3.将15个温度指标和13个降水量指标进行主成分分析得其主成分，并以2010年市县疟疾发病率为因变量，男性人口、农业人口、少数名族人口、GDP、第一产业、第二产业、第三产业、人均GDP、卫生机构总数、执业医师人数、农村居民家庭人均纯收入、城市居民家庭人均纯收入、温度主成分、降雨量主成分、湿度等为解释变量构建结局变量为连续型变量的多元线性回归模型。其中，疟疾发病的危险因素的包括“城市居民家庭人均纯收入”、“降水量主成分2”、“湿度”这三个因素，疟疾发病的保护因素包括“男性人口”、“农业人口”、“人均GDP”、“医疗机构个数”、“执业医师个数”、“农村居民家庭人均纯收入”这6个因素。研究结论本课题重点探讨了1990-2010年海南省疟疾流行病学特征和时空聚集性，还重点研究了海南省不同市县疟疾疫情存在差异的主要影响因素。在海南，疟疾的流行有一定的周期性、地域性，并在某些特定人群中有较高的发病率。时间序列模型可以很好的拟合疟疾发病率在时间序列上的变化趋势。在防治措施、人口免疫状态及人口流动没有发生大幅度变化的时候可以用来预测疟疾的发病率变化。海南省2010年起开始实施的消除疟疾计划作用明显，时间序列模型从侧面也证明了这一点。另外，降水量和湿度将影响海南省疟疾在各市县的发病，而经济、卫生和人民生活水平也是影响疟疾发病的重要因素。这一研究结果将有助于当地进一步在多方面对疟疾疫情进行防控。创新点本课题在整个研究过程中，将多个领域的信息与知识进行了整合。将疟疾疫情数据进行时间、空间上的整合，打破了时间与空间的限制，不仅仅研究了有网络直报系统以来的数据，而且还研究了从1990年来的疟疾发病数据。在空间上，打破了县市的界限，对海南省全省疫情进行空间扫描，为疟疾疾病的预警提供新思路。将人口学、经济学及卫生学知识相融合，使得疟疾疫情及影响因素的分析更加的立体，更加的贴合实际。将自然因素与社会因素相结合的分析方法在国内外的疟疾研究领域较为罕见。更多还原

【Abstract】 ObsjectivesThe study explored the spatio-temporal characters of malaria and the factors effectedthe incidence of malaria in Hainan province since1990to2010in order to provide theevidence for the prevention and control policy of malaria in Hainan and other similardistricts. It also provided a methodological reference for similar studies by thecomprehensive application of time and space methods in the field of malaria research.Results1. The incidence of malaria was decreased in the past few years in Hainan Province. Overall,the highest peak of malaria infection was July and August was the minimum peak. Hainanwas February. The southwest region in Hainan is the area of highest incidence of malariadisease. Population distribution,30-50-year-old male resident was reported to be high-riskpopulation. We found that peasants, workers and migrant workers accounting for85.11%ofthe total number of reports.2. We applied space-time permutation scan statistic to analyze malaria incidence data andfound that from2005to2010, there was seven malaria gathering area in Hainan (P <0.05).For malaria incidence data in2010, we found three gathering area (P <0.05). using data fromJanuary2005to December2009to establish ARIMA model to predict the incidence of thedisease in January2010, the predictive value was0.15/10million，95%confidence intervalwas [-1.04,1.33]. The actual detection incidence rate of this month was0.17/10million, andthe actual monthly incidence value falls within the95%confidence interval of the predictedvalue, the relative prediction error was11.8%. We re-fitting the model after the actual incidence of malaria in January2010included in the time series model to predict theincidence of malaria from February to December in2010, the actual incidence of malariawas85.75%lower than the theoretical incidence of malaria.3. We obtained the principal component from15temperature indicators and13precipitationindicators by using principal component analysis. The dependent variable of multiple linearregression model was the incidence of malaria in2010, and the outcome variables were themale population, agricultural population, minority ethnic population, GDP, first industry,secondary industry and tertiary industry, GDP per capita, the total number of healthinstitutions, the number of practicing physicians, net income of rural households per capita,net income of urban households per capita, the main component of the temperature, the maincomponent of the rainfall, humidity and other explanatory variables. The risk factorsincluding net income of urban households per capita, the main component of theprecipitation and humidity, and the protective factors include the male population,agriculture population, per capita GDP, the number of medical institutions, the number ofpracticing physicians, net income of rural households per capita.ConclusionThis study focused on the malaria epidemiological characteristics and spatio-temporalaggregation since1990in Hainan Province, and we also studied the main factors associatedwith different incidents of malaria among different cities and counties in Hainan. Themalaria in Hainan was infected with a certain periodicity, in a certain regional, and with ahigher incidence in certain populations. Time series model can be good fit to the trend of theincidence of malaria in the time series. It can be used to predict changes in the incidence ofmalaria under the premise of prevention and control measures, the population of immunestatus and population movements did not significantly changed. The policy and instructionimplemented since2010to eliminate malaria program in Hainan is obvious, the forecastresult of time series model in2010also proved this point. In addition, precipitation andhumidity will affect the incidence of malaria in different cities and counties. And economic,health and people’s living standards are also important factors associated with the incidenceof malaria. The results of this study will help the local relevant departments furtherstrengthen the prevention and control of malaria epidemics.Innovation This study integrated a number of areas of information and knowledge. The integrationof time and space for the malaria epidemic data broke the limitations of time and spaceresearch. It not only analyzed the data since direct reporting network system developed, butalso study the malaria incidence data from1990. In space, our study broke the boundaries ofthe counties of Hainan Province. We use space scanning scaned the epidemic situation of thewhole province to provide new ideas for early warning of malaria disease. The integration ofdemography, economics, and health knowledge made malaria epidemic and its influencingfactors analysis more dimensional, more fit the actual. The analysis method which combinednatural factors and social factors in the field of malaria research is rare at home and abroad.更多还原