【作者】 王璐；

【导师】 宇传华；

【作者基本信息】 武汉大学 ， 统计学， 2019， 博士

【摘要】 目的:道路交通伤害(Road Traffic Injuries,RTIs)一直是中国重要的公共卫生问题。本研究通过对1990-2017年中国RTIs死亡率数据的描述和定量分析,了解中国RTIs的死亡率变化趋势。利用中国RTIs死亡数据拟合年龄-时期-队列模型、Smeed’s模型、Borsos’s模型和空间自相关,分析中国RTIs死亡趋势的流行病学特征,并评估ARIMA模型、分数多项式与自然立方样条回归模型在RTIs死亡率中的预测效用,为相关部门制定RTIs的防控策略或措施提供科学依据。方法:本研究以中国人群为研究对象,分别收集整理1990-2017年中国RTIs的死亡相关数据。主要对以下五个方面进行分析研究:(1)采用Joinpoint回归分析模型,对我国1990-2017年不同性别和道路不同使用者的RTIs整体死亡趋势进行研究,并估计其趋势变化的年度百分比变化和平均年度百分比变化及其95%可信区间。(2)描述1990-2017年中国不同性别RTIs死亡率的年龄、时期和队列趋势,并采用内生因子法算法的年龄-时期-队列模型,评估年龄、时期与队列对RTIs死亡率的独立影响;(3)运用Smeed’s模型和Borsos’s模型,分析1996-2107年间中国RTIs死亡率随机动化率的变化情况,并比较这两种模型的拟合效果;(4)运用全局自相关和热点分析,从空间层面上描述1996-2017年中国RTIs死亡的空间分布;(5)对1990-2017年中国不同性别RTIs标化死亡率数据构建ARIMA模型、分数多项式和自然立方样条回归模型,并利用成功构建的模型对中国2018-2022年不同性别RTIs年龄标化死亡率进行预测。结果:(1)在1990-2017年间,中国RTIs年龄标化死亡率整体呈现先上升后下降的趋势,峰值在2000年,2011年之后下降速度明显加快。Joinpoint结果显示,全人群RTIs死亡率有三段有意义的趋势变化:1990-2000、2000-2011和2011-2017年,1990-2000年间死亡率的以每年1.3%的速度上升,2000-2011和2011-2017年间分别以每年1.6%和2.9%的速度下降。1990-2017年间,男性RTIs的死亡率一直明显高于女性。中国男性RTIs死亡率有意义的变化趋势与全人群一致,趋势分段也是1990-2000、2000-2011年和2011-2017年。1990-2000年死亡率以每年1.5%的速度上升,2011-2017年,1990-2000年间分别以每年1.4%和2.9%的速度下降。在1990-2017年间,中国女性RTIs的标化死亡率经历了4段有意义的趋势变化。1990-1996年死亡率以每年1.5%的速度上升,2001-2007、2010-2015和2015-2017年间分别以每年2.5%、2.5%和4.8%的速度下降。道路不同使用者的RTIs死亡率在全人群、男性和女性的趋势大体一致,呈现先升后降趋势。不同类型道路使用者中,行人RTIs死亡率均是最高,其后在男性中依次是摩托车、机动车、自行车和其他;而在全人群与女性中依次是机动车、摩托车、自行车和其他。此外,1990-2017年间,骑自行车人的RTIs死亡率是以平均每年1.1%的速度上升的,其余类型均是下降的。(2)随着年龄的增加,我国不同性别RTIs死亡率基本呈现上升趋势,在90-94岁组达到最高;总体死亡率随出生年代呈现下降趋势。采用内生因子算法构建年龄-时期-队列模型,分解1990-2017年中国不同性别RTIs死亡率的年龄、时期与队列效应。研究结果显示,随着年龄的变化,男性与女性RTIs的死亡风险的趋势变化大致相同。RTIs死亡率随年龄增加整体呈现逐渐上升的趋势,随队列增加则表现为逐渐下降的趋势。此外,消除年龄与队列两个因素的影响后,时期效应增加了男性RTIs的死亡风险,而降低了女性RTIs的死亡风险。男性与女性RTIs死亡风险的队列效应趋势基本一致,出生在1953-1957的人均有着最高的死亡风险,当消除年龄与时期这两个因素的影响后,男性与女性RTIs的死亡率的队列效应较之前均有所升高。(3)机动化水平随着人均GDP上升呈现逐年上升趋势。从1996到2017年,随着机动化水平的升高,十万车死亡率则逐渐下降,十万人死亡率呈现先升后降的趋势。整个观察期内,十万车死亡率和十万人死亡率整体分别下降了91.28%和23.77%,十万车死亡率降速度更快。随着机动化率的升高,RTIs死亡率的Smeed’s曲线和Borsos’s曲线均呈现下降趋势,但Borsos’s曲线更贴近死亡率观察值,因此Borsos’s模型能更好的拟合中国RTIs的死亡数据。(4)1996年中国RTIs死亡率主要分布在西部的新疆、西藏、等地,以及沿海的浙江和江苏等地。与1996相比,2017年以上地区RTIs死亡率有所降低,但湖北、贵州和吉林等地成为RTIs高发地区。2007-2017年间RTIs的死亡率下降明显快于1996-2006年。全局自相关分析显示中国RTIs死亡率是空间随机分布,无聚类或离散趋势。热点分析结果显示RTIs热点区域与1996-2017年空间死亡率变化相似。(5)对1990-2017年中国不同性别的RTIs标化死亡率数据建立ARIMA模型、分数多项式与自然立方样条回归模型。经过对男性与女性RTIs死亡率数据序列进行平稳性和随机性检验后,终止ARIMA预测模型。因此建立FP(-1-1)和NCS回归模型,拟合中国男性与女性RTIs死亡率,并对不同模型的拟合效果进行评价,确定NCS回归模型为最优模型。根据NCS回归模型的预测结果可知,未来五年中国男性与女性RTIs死亡率依旧会持续下降,男女差距可能会缩小。依照现在的下降速度,中国可能无法完成联合国可持续发展目标至2020年RTIs死亡数减少一半的目标。结论:(1)在1990-2017年间,我国RTIs死亡率水平总体上呈先升高后降低的趋势,且男性高于女性人群。行人、骑自行车和骑摩托车的人是RTIs的弱势人群。(2)当消除时期与队列因素影响后,中国男性与女性在15-29岁和60岁以上年龄组的RTIs死亡风险均随着年龄的增加而增加。当消除年龄与队列这两个因素的影响后,时期效应增加了男性RTIs的死亡风险,降低了女性的RTIs的死亡风险。当消除年龄与时期这两个因素的影响后,队列效应增加了男性女性的RTIs的死亡风险。(3)随着经济与机动化水平升高,死亡率呈现下降趋势。Borsos’s模型的拟合RTIs死亡的效果更好。(4)中国RTIs死亡率空间上是随机分布的,且不可忽视湖北、贵州和吉林等非高发地区RTIs的防控。(5)未来五年中国RTIs死亡率会持续下降,男女差距进一步缩小,但中国可能无法完成来联合国至2020年RTIs死亡人数减少一半的目标。更多还原

【Abstract】 Objectives:Road traffic injuries(RTIs)have always been an important public health issue in China.In this study according to the data of RTIs mortality in China from 1990 to 2017,we make a fully understanding of the trends of RTIs mortality among different genders and road users in China in the past 30 years.In addition,the epidemiological characteristics of RTIs mortality were analyzed by setting up the age-period-cohort model,Smeed’s model,Borsos’ s model and spatial auto-correlation analysis.Meanwhile,the ARIMA model,fractional polynomial regression model(FP)and natural cubic spline regression model(NCS)were applied to explore the prediction of RTIs mortality rate,which provides a scientific basis for relevant departments for the formulation of prevention and control strategies on the RTIs and lays a foundation for evaluating the effects of RTIs prevention in the future.Methods:In this study,the Chinese population was selected as the research object,and the death-related data of RTIs from 1990 to 2017 were collected.The contents mainly include the following five aspects:(1)We studied the trends of of RTIs mortality rate among different genders and types of road users in China from 1990 to 2017 by using the Joinpoint regression model,and estimated the annual percent,the average annual percentage change and its 95% confidence interval of the trend changes.(2)The age,period and cohort trends of RTIs mortality in different genders in China from 1990 to 2017 were described,and the age-period-cohort model with the Intrinsic Estimator was used to assess the independent age,period and cohort effects on RTIs mortality.(3)We analyzed the trend changes of RTIs mortality with the motorization rate from 1996 to 2107 by the Smeed’s model and Borsos’ s model,and compared the fitting effects of the two models.(4)We described the spatial distribution of RTIs mortality in 1996-2017 by the spatial auto-correlation and hot spot analysis.(5)We constructed the ARIMA model,FP regression model and NCS regression model for RTIs mortality in different genders in China from 1990 to 2017,and predicted the age-standardized mortality of RTIs from 2018 to 2022 by the successfully constructed model.Results:(1)The RTIs mortality rate in China showed a tendency of increase first and then decrease from 1990 to 2017 with the highest value in 2000,and the decline rate was significantly accelerated after 2011.The results of Joinpoint showed that the trend RTIs mortality had three significant segments: 1990-2000,2000-2011 and 2011-2017.The mortality rate in 1990-2000 increased at a rate of 1.3% per year,and decreased at a rate of 1.6% and 2.9% per year in 2000-2011 and 2011-2017,respectively.The mortality rates of RTIs in males from 1990-2017 were significantly higher than that in females.The significant segments of RTIs mortality in male were consistent with that in the whole population.The mortality rate during the segment 1990-2000 increased at a rate of 1.5% per year,and during the segments 2011-2017 and 1990-2000 decreased at a rate of 1.4% and 2.9% per year,respectively.The RTIs mortality rate in female experienced four significant trends between 1990 and 2017.The mortality increased at the rate of 1.5% per year in 1990-1996,and during the segments 2001-2007,2010-2015 and 2015-2017 decreased at the rate of 2.5%,2.5% and 4.8% per year,respectively.The trends RTIs mortality among different types of road users were generally consistent with that in the whole population,male and female,showing a tendency of increase first and then decrease.The pedestrian mortality rate was highest among different types of road users,followed in male by motorcyclist,motor vehicles,cyclists and others,while in the whole population and female by motor vehicles,motorcyclist,cyclists and others.Joinpoint analysis showed the similar trend among different types of road users,but the significant segments were different between 1990 and 2017.In addition,the RTIs mortality rate for cyclists increased at an average rate of 1.1% per year.(2)The mortality rate of RTIs in different genders increased with age,reaching the highest in the 90-94 age group,and showed a downward trend with the birth cohort.The independent effects of age,period,and cohort for RTIs mortality were separated by the age-period-cohort model with IE algorithm.Moreover,the risk of RTIs mortality in different genders were roughly similar with age.We found that the overall risk of RTIs mortality in male and female increased with age and decreased with the birth cohort.In addition,when controlling the age and cohort effects,the period effect increased the risk of RTIs mortality in male and decreased the risk of RTIs mortality in female.The cohort effects of RTIs mortality between male and female were also roughly similar.The people born in the 1953-1957 had the highest risk of death,and when controlling the age and period effects,the cohort effect increased the risk of RTIs mortality in both male and female.(3)The level of motorization showed an upward trend with the increase of GDP per capita.From 1996 to 2017,with the increase of motorization level,the RTIs death per100,000 vehicles gradually decreased,and the RTIs death per 100,000 population showed a tendency of increase first and then decrease.Throughout the observation period,the RTIs death per 100,000 vehicles and RTIs death per 100,000 population decreased by 91.28% and 23.77%,respectively,and the RTIs death per 100,000 vehicles decreased faster.With the increase of motorization rate,the Smeed’s curve and Borsos’ s curve of RTIs mortality rate showed a downward trend,but the Borsos’ s curve was closer to the mortality observed values.So Borsos’ s model could better fitting the death of RTIs in China.(4)In 1996,the mortality rate of RTIs in China was mainly distributed in the west regions,such as Xinjiang and Tibet etc.,as well as in coastal regions,such as Zhejiang and Jiangsu and so on.Compared with 1996,the mortality rate of RTIs in the above areas decreased in 2017,but the regions,such as Hubei,Guizhou and Jilin has been the high-risk areas of RTIs mortality.The decline in mortality of RTIs between 2007 and 2017 was significantly faster than the that between 1996 and 2006.The spatial autocorrelation analysis showed that RTIs mortality rate in China was spatially randomized,with no clustering or discrete trend,and the hot-spot analysis results showed that the hot-spot areas were similar to the spatial changes of RTIs mortality from 1996 to 2017.(5)This study constructed the ARIMA,FP and NCS regression models by using the data of RTIs mortality in different genders in China from 1990 to 2017.The ARIMA model was terminated after a stationary and randomized detection of the RTIs mortality in different genders.So the FP(-1-1)regression and the NCS model were established for RTIs mortality in different genders.The fitting effects of different models were evaluated,and the NCS regression model is finally determined as the optimal model.According to the prediction results of the NCS regression model,the RTIs mortality in male and female will continue to decline in the next five years,and the gap between male and female may shrink.According to the current rate of decline,China may not be able to achieve the UN Sustainable Development Goal-halving the number of RTIs deaths by 2020.Conclusions:(1)Between 1990 and 2017,the RTIs mortality showed showed a tendency of increase first and then decrease,and the mortality in male was higher than that of female.Pedestrians,cyclists and motorcyclists are vulnerable to RTIs.(2)When controlling the period and cohort effects,the RTIs mortality in male and female increased with age at the age group of 15-29 years and over 60 years.when controlling the age and cohort effects,the period effect increased the risk of RTIs in male and reduced the risk of RTIs mortality in female.when controlling the age and period effects,the cohort effect increased the risk of RTIs mortality in male and female.(3)With the increase of economy and motorization levels,the RTIs mortality rate showed a downward trend.Borsos’ s model had better effects on fitting the RTIs mortality.(4)The mortality rate of RTIs in China was randomly distributed,and the prevention and control of RTIs in non-high-incidence areas such as Hubei,Guizhou and Jilin could not be ignored.(5)RTIs mortality rate in China would continue to decline in the next five years,and the gap between male and female would further narrow,but it is almost impossible for China to complete the UN goal of halving the number of RTIs deaths by 2020.更多还原