【作者】 王海鹏；

【导师】 孟庆跃；

【作者基本信息】 山东大学 ， 社会医学与卫生事业管理， 2013， 博士

【摘要】 研究背景随着社会经济的迅速发展,生活水平的提高,生活模式的转变,以及人口老龄化的快速到来,全球糖尿病患病率和患者数正以惊人的速度增长。糖尿病及其并发症不仅给患者的身心健康造成严重伤害,也给患者、家庭和社会带来沉重的经济负担,是目前世界上最普遍和最具挑战性的重大公共卫生问题之一。在我国,近20年来糖尿病患病率增长了10倍左右,导致糖尿病疾病经济负担的大幅度增加。根据世界银行亚太地区报告的预测,糖尿病将在下一个十年成为我国最流行的疾病。糖尿病患者数的快速增加,必将需要更多的卫生服务和资源投入,这给国家卫生规划和资源配置带来新的挑战。然而,目前卫生政策制定通常是根据以前的或者现有的数据来计划卫生服务和资源投入,未考虑未来疾病谱和经济负担的长远变化趋势,这种基于实际需求而非未来发展趋势的估计会低估需求资源的投入。因此,根据历史趋势对糖尿病疾病经济负担的未来发展趋势进行科学预测,是值得深入探讨的课题,也是亟需解决的政策问题。我国糖尿病疾病经济负担研究起步较晚,虽然有不少学者陆续开展了相关研究,但这些研究比较零散,研究地区、内容和方法并不一致,研究结果也有较大差异。由于我国卫生信息系统尚不完善,糖尿病相关基础数据资料缺失,导致国家层面的糖尿病疾病经济负担研究较少。在我国糖尿病疾病经济负担预测研究方面,能够查阅到的文献更少,且预测方法较粗,研究时间较早,难以起到服务于未来卫生政策制定的作用。因此,亟需借鉴发达国家的研究经验和方法,探索糖尿病疾病经济负担预测理论和研究思路,构建我国糖尿病疾病经济负担预测模型,将预测研究结果应用于我国卫生研究和卫生规划实践,服务于我国卫生政策制定和资源规划。本论文将系统总结国内外糖尿病疾病经济负担预测相关的理论和模型,在此基础上构建本研究的糖尿病疾病经济负担预测模型,预测我国未来糖尿病疾病经济负担,并探索其增长的驱动因素,为我国卫生规划和资源配置提供科学依据,从而实现“以需要为基础”的糖尿病卫生服务规划与提供,以确保资源投入能够匹配未来需求。基于此,本论文提出的研究问题是：如何预测我国糖尿病疾病经济负担的未来发展趋势。理论问题是,如何借鉴目前糖尿病疾病经济负担预测理论,构建我国糖尿病疾病经济负担预测模型?实证研究中,如何根据历史趋势来预测我国诊断糖尿病疾病经济负担的未来发展趋势及其驱动因素?研究目的本论文的总目标是通过理论模型研究和实证分析,探讨如何构建我国糖尿病疾病经济负担预测模型,进而根据我国诊断糖尿病疾病经济负担的历史趋势,来预测我国诊断糖尿病疾病经济负担的未来发展,并分析其发展变化的推动因素,为我国卫生规划和资源配置提供数据基础和科学依据。具体目标是：系统总结国内外糖尿病疾病经济负担预测研究相关的理论模型；构建我国糖尿病患病预测模型和疾病经济负担预测模型；揭示1997-2009年我国诊断糖尿病疾病经济负担的历史趋势及其增长动因;预测2010-2020年我国诊断糖尿病疾病经济负担的未来发展及其驱动因素；提出符合未来需求的我国卫生规划和资源配置政策建议。研究方法糖尿病疾病经济负担预测研究的思路是依据历史趋势分别对我国人口数、糖尿病患病率和人均疾病经济负担的未来趋势进行预测,将三者的乘积作为相应年份总的糖尿病疾病经济负担；预测研究采用基于组份模型,按照年龄分组分别进行预测和计算,然后加总得到未来总的糖尿病疾病经济负担。糖尿病疾病经济负担主要包括疾病负担、直接经济负担和间接经济负担。在系统总结国内外已有糖尿病疾病经济负担预测模型的基础上,首先构建糖尿病群体患病率预测模型和个体患病概率预测模型,糖尿病患病率预测采用趋势外推法和数据函数模型,糖尿病患病概率影响因素分析采用面板数据回归模型；然后,构建人均疾病经济负担预测模型和医疗服务利用及费用预测模型,糖尿病患者人均医疗费用预测采用趋势外推法和年增长率法,医疗服务利用及费用影响因素分析采用样本选择模型。在糖尿病疾病经济负担的评价指标和测算方法上,采用两个指标来衡量糖尿病患者的疾病经济负担情况,一是因所有疾病造成的疾病经济负担,二是糖尿病给患者造成的额外疾病经济负担；疾病负担采用患者数和因病误工天数等指标来评价,经济负担测算采用疾病成本法,直接经济负担用医疗费用来评价,间接经济负担采用人力资本法来测算。最后,采用增长因素分解法中的拉氏分解法来分解各因素对糖尿病疾病经济负担增长的贡献度。本研究的主体数据来源于北卡罗来纳大学和中国疾病控制中心联合进行的长期纵向国际合作项目——中国健康和营养调查(CHNS)。该项目考虑不同经济发展水平和膳食结构,选择了辽宁、黑龙江、江苏、山东、河南、湖北、湖南、广西、和贵州等9个省份作为样本地区,通过多阶段分层随机抽样确定了大约4400个家庭,涉及到约19000个调查对象,该项目每隔2-4年对样本人群队列进行一次追踪调查。由于CHNS中的糖尿病调查从1997年开始,因此我们选用了1997年、2000年、2004年、2006年、2009年等5个年份的数据。本论文以五轮调查中所有20岁及以上成年人为研究对象,同一调查对象有1-5条记录,共有48696条记录,本研究的糖尿病患者为诊断糖尿病患者。另外,人口数据来自美国人口统计局所做的中国人口预测数据。主要研究内容包括诊断糖尿病患病率和医疗费用及其影响因素、1997-2009年我国诊断糖尿病疾病经济负担的历史趋势、2010-2020年我国诊断糖尿病疾病经济负担的未来发展。所采用的指标变量主要包括健康状况及健康相关危险因素、社会人口经济学特征、医疗服务利用及疾病经济负担等。研究结果1.研究对象的基本情况。本研究以20岁及以上人群为研究对象,从1997年到2009年,女性比例约为52%,随年份变化不大；20-39岁组所占比例下降了约20个百分点,40-59岁组和60岁及以上组分别升高了约10个百分点；小学及以下组所占比例下降了约8个百分点,初中组、高中组和大学组所占比例均升高了约3个百分点；无工作者所占比例升高了17个百分点,退休者升高了5个百分点；城镇户籍所占比例约为33.3%,随年份变化不大；医疗保险覆盖率上升了约64个百分点；经居民消费价格指数调整后,家庭人均收入从4013元增长到11487元。与非糖尿病人群相比,糖尿病患者人群的年龄结构较为老化,整体学历水平较低,无工作者和退休者所占比例较高,城镇户籍所占比例较高。总人群四周患病率增长了约8个百分点,慢性病患病率增长了约9个百分点,肥胖率和超重率分别增长了约5个百分点和14个百分点,测量高血压率增长了约7个百分点,中重度体力活动率下降了将近15个百分点,吸烟率和饮酒率均降低了约4个百分点。比较来看,糖尿病人群的四周患病率、肥胖率、超重率和测量高血压率均明显高于非糖尿病人群；但是,糖尿病人群的中重度体力活动率、吸烟率、饮酒率等则低于非糖尿病人群。2.糖尿病患病率和医疗费用的影响因素。不同特征分组人群之间糖尿病患病率的差异性分析结果显示,不同年龄、学历、职业、收入、户籍等分组的糖尿病患病率之间的差异具有统计学意义(P<0.05),60岁及以上年龄组、小学及以下学历组和大学及以上学历组、退休者和无工作者组、低收入和高收入组、城镇户籍组等的糖尿病患病率较高；不同体重指数、血压、体力活动、吸烟、饮酒等特征分组的糖尿病患病率之间的差异也具有统计学意义(P<0.05),超重组为轻或正常组的2倍多,肥胖组为轻或正常组的4倍多,高血压组约为非高血压组的3倍,非中重度体力活动组为中重度体力活动组的3倍多。糖尿病患病率的多因素分析结果发现,年龄、肥胖、高血压和中重度体力活动等因素对糖尿病患病率的影响均具有统计学上的显著性(P<0.05),60岁以上组和40-59岁组患糖尿病的概率比20-39岁组更高,肥胖者患糖尿病的概率比非肥胖者更高,高血压者比非高血压者更容易患糖尿病,中重度体力活动者患糖尿病的概率比非中重度体力活动者更低。例均医疗费用的差异性分析结果显示,不同性别、年龄、学历、职业、收入、户籍、医疗保险分组例均医疗费用之间的差异具有统计学意义(P<0.05),男性约为女性的1.37倍,60岁及以上组约为20-39岁组的1.53倍,退休者组和无工作者组的例均医疗费用较高,高收入组约为低收入组的1.88倍,城镇户籍约为农村户籍的1.65倍,城职保组和公费劳保医疗组为无医疗保险组的2倍左右。但是,不同性别、年龄、学历、职业、家庭人均收入、户籍和医疗保险分组的糖尿病患者之间例均医疗费用的差异均无统计学意义(P>0.05)。医疗费用的多因素分析结果证明,年龄、婚姻状况、文化程度、职业、医疗保险、家庭人均年收入、户籍和是否住院等对医疗费用的影响均具有统计学意义(P<0.05),高年龄组的医疗费用比低年龄组高,已婚居民的医疗费用比其他婚姻状况组的高,居民医疗费用随着教育年限和家庭人均收入的增加而升高,而有工作居民的医疗费用比无工作和退休者组的低,医疗保险则能够增加居民的医疗费用,有住院服务利用的居民医疗费用明显较高,农村居民的医疗费用低于城镇居民。3.我国诊断糖尿病疾病经济负担的历史趋势。从1997年到2009年,我国20岁以上人群诊断糖尿病患病率从0.95%增长到了2.24%,20-39岁组、40-59岁组和60岁及以上组分别增长了0.07、1.11和4.15个百分点；诊断糖尿病患者数从770.17万增加到了2159.59万,20-39岁组、40-59岁组、60岁及以上组分别增加了30.58万、569..93万和786.91万；诊断糖尿病患者数的增长中,人口变化引致的增加为265.30万,患病率升高引致的增加为807.41万,二者共同作用引致的增加为314.71万。我国诊断糖尿病患者总因病误工天数从97.08百万天增长到了372.73百万天,总增长率为283.94%其中因患者数增加引致的增长率为199.94%,人均因病误工天数增加引致的增长率为28.94%二者共同作用引致的增长率为55.06%;额外因病误工天数从41.91百万天增长到了233.53百万天,总增长率为457.14%,其中因患者数增加引致的增长率为179.20%,因人均额外因病误工天数增加引致的增长率93.38%,二者共同作用引致的增长率为184.56%。我国诊断糖尿病患者的总医疗费用从247.19亿元增长到978.93亿元,总增长率为296.02%,其中因患者数增加引致的增长率为190.79%,因人均医疗费用增加引致的增长率为36.64%,二者共同作用引致的增长率为68.59%;额外医疗费用从197.19亿元增长到682.95亿元,总增长率为246.34%,其中因患者数增加引致的增长率为180.73%,因人均额外医疗费用增加引致的增长率为25.16%,二者共同作用引致的增长率为40.45%。我国诊断糖尿病患者的总间接经济负担从20.07亿元增加到了261.17亿元,总增长率为1201.53%,其中因患者数增加引致的增长率为199.47%,因人均间接经济负担增加引致的增长率为340.49%,二者共同作用引致的增长率为661.57%,额外间接经济负担从8.66亿元增加到了163.63亿元,总增长率为1498.90%,其中因患者数增加引致的增长率为179.26%,因人均额外间接经济负担增加引致的增长率为563.73%,二者共同作用引致的增长率为1045.91%o4.我国诊断糖尿病疾病经济负担的未来预测。从2010年到2020年,我国20岁以上人群诊断糖尿病患病率将从2.313%升高到4.664%,其中20-39岁组、40-59岁组和60岁及以上组分别上升0.018、1.673和5.815个百分点；诊断糖尿病患者数将从2285.55万增加到5013.56万,其中20-39岁组将减少3.36万,40-59岁组将增加802.66万,60岁及以上组将增加1928.41万；诊断糖尿病患者的总增长率将为119.35%,其中因人口变化引致的增长率为25.86%,因患病率升高引致的增长率为71.79%,二者共同作用引致的增长率为21.70%。我国诊断糖尿病患者的总因病误工天数将从403.63百万天增长到1084.51百万天,总增长率为168.69%,其中因患者数增加引致的增长率为128.63%,由人均因病误工天数增加引致的增长率为19.74%,二者共同作用引致的增长率为22.12%;额外因病误工天数将从256.27百万天增长到773.79百万天,总增长率为201.95%,其中因患者数增加引致的增长率为123.51%由人均额外因病误工天数增加引致的增长率为33.96%,二者共同作用引致的增长率为44.48%。我国诊断糖尿病患者的医疗费用预计将从1094.15亿元增长到4020.53亿元,总增长率为267.46%,其中因患者数增加引致的增长率为125.11%,因人均医疗费用增加引致的增长率为63.47%,二者共同作用引致的增长率为78.88%;额外医疗费用将从757.07亿元增长到2541.65亿元,总增长率235.72%,其中因患者数增加引致的增长率为117.39%因人均医疗费用增加引致的增长率为55.34%,二者共同作用引致的增长率为62.99%。我国诊断糖尿病患者的间接经济负担将从305.44亿元增长到1691.27亿元,总增长率为453.71%,其中因患者数增加引致的增长率为128.63%,因人均间接经济负担增加引致的增长率为143.06%,二者共同作用引致的增长率为182.02%额外间接经济负担将从193.93亿元增长到1206.72亿元,总增长率将为522.25%,其中因患者数增加引致的增长率为123.45%,因人均额外间接经济负担增加引致的增长率为176.02%,二者共同作用引致的增长率为222.78%。结论和政策建议我国诊断糖尿病疾病经济负担预测研究对于满足未来需求的卫生规划和资源配置有着至关重要的意义。预测的重点和难点是预测模型构建和数据获取,这是目前国内糖尿病疾病经济负担预测研究较少的主要原因。本研究在系统总结国内外糖尿病疾病经济负担预测模型的基础上,对我国未来年份的诊断糖尿病疾病经济负担开展了量化预测。从糖尿病疾病经济负担的流行和分布情况来看,我国正处于糖尿病患病率快速增长的经济社会转型时期,经济发展水平越高的省份糖尿病患病率越高,城市患病率高于农村,同一地区随着经济发展水平的提高糖尿病患病率越来越高。糖尿病患病率在不同特征人群中存在一定程度的差异,高年龄、低学历、无工作者、低收入等弱势人群的糖尿病患病率较高。医疗费用在不同特征分组人群中也存在一定程度的差异,高年龄和无工作者等弱势人群的医疗费用也较高；然而,不同特征分组糖尿病患者之间医疗费用的差异不大,说明有糖尿病对所有特征人群均意味着沉重的医疗费用负担。与非糖尿病居民相比,糖尿病患者的医疗服务利用率和次均费用均较高,故其人均年医疗费用明显更高；糖尿病患者的医疗费用明显较高,而人均年收入和家庭年收入较低,所以个人经济负担和家庭经济负担均明显高于非糖尿病居民。我国诊断糖尿病疾病经济负担的流行已经给人群健康和社会经济带来了沉重的负担。由于糖尿病危险因素、医疗费用和经济发展水平的持续快速升高,以及人口老龄化的到来,我国诊断糖尿病患病率和患者数将迅猛增长。如不采用有效的控制措施,我们诊断糖尿病疾病经济负担将继续快速增长。从我国诊断糖尿病疾病经济负担的增长驱动因素来看,诊断糖尿病患病率增长对诊断糖尿病患者数增加的推动作用大于人口变化的影响,诊断糖尿病患者数增加对我国诊断糖尿病疾病经济负担增长的贡献大于诊断糖尿病患者人均疾病经济负担的作用,但人均疾病经济负担的增长也不容忽视。根据本研究的结论,提出以下政策建议：(1)为更好地对我国未来的健康问题进行预测研究,应加强数据资料的收集工作,进一步完善各级卫生部门的电子信息系统；(2)依据预测的我国诊断糖尿病疾病经济负担未来流行趋势及其增长动因,开展满足未来需求的糖尿病相关卫生规划和资源配置；(3)为了控制和延缓我国糖尿病疾病经济负担的增长,应针对可以控制的因素采取有效的预防干预措施,比如控制饮食、降低肥胖和高血压患病率等,并加强糖尿病患者的管理,从源头上减少糖尿病疾病经济负担；(4)对于糖尿病疾病经济负担比较沉重的重点人群,需要制定特殊的卫生政策和援助计划,如将糖尿病门诊服务纳入医疗保险支付,扩大医疗保险对糖尿病医疗费用的报销力度,以切实减轻糖尿病患者的疾病经济负担,提高糖尿病患者及其家庭的生活质量。创新与不足本研究创新性：(1)目前尚未发现关于糖尿病疾病经济负担预测模型的系统研究,不同研究采用的预测模型各异,得到的预测结果也存在很大差异。本论文在文献综述的基础上,系统总结国内外糖尿病疾病经济负担预测模型,比较分析其优缺点,并借鉴数理统计学和计量经济学的思路和方法,对现有糖尿病疾病经济负担预测模型进行调整和完善。(2)目前国内缺少糖尿病疾病经济负担预测方面的研究。本论文在理论模型研究的基础上,对我国未来的糖尿病疾病经济负担开展了预测研究,并探讨预测过程中会遇到的问题和解决办法,弥补国内该领域研究的不足,也为其他健康问题的预测研究奠定基础。(3)采用面板数据模型和样本选择模型研究糖尿病患病和医疗费用的影响因素。模型本身所具有的优点,使得研究结论更加可靠,有助于依据影响因素的变化趋势来判断糖尿病疾病经济负担的未来发展。首次采用因素分解法对未来糖尿病疾病经济负担的增长进行分解,有利于为未来政策制定找到干预点。本研究不足及展望：(1)本论文预测研究主要采用趋势外推法,对糖尿病患病率的预测采用数学函数模型,对人均疾病经济负担的预测采用年增长率法,把所有影响因素的作用归结到时间变量上,未把各种影响因素变量作为参数纳入到预测模型中。以后的研究应该纳入这些参数,向宏观面板数据模型、微观模拟模型和疾病长期费用模型方向发展。(2)本研究所利用的数据资料年份较少,这会对预测结果的准确性造成一定程度的影响。将来可以长期追踪收集数据资料,把新获得的资料加入到时间序列数据中,重新拟合模型来对其进行校正和数据更新。另外,有很多关系复杂的因素影响糖尿病患病率和医疗费用,社会经济政策环境的未来变化会使原有模型的预测效果下降,未来研究可以考虑动态情景分析法。更多还原

【Abstract】 BackgroundWith the rapid development of society and economy, the improvement of living standard, change of lifestyle and rapid aging of population, the global prevalence and patients of diabetes are rising at an alarming pace. As diabetes and its complications not only cause serious injury to physical and mental health of patients, but also make a heavy burden to patients, families and society. It’s one the most popular and most challenging public health problems. In China, the prevalence of diabetes increased by10times in recent20years, and the economic burden of diabetes had been increasing rapidly. Predicted by the Word Bank’s Asia-Pacific reports, diabetes will become the most prevalent disease of next decade in China. The rapid growth of diabetes patients calls for more health services and resources, and it brings new challenges for national health planning and resource allocation. Health policy is usually made based on previous or existing data, however, the estimate of existing demand is supposed to underestimate the actual situation in the future. Therefore, making scientific prediction on diabetes’economic burden by historical trends analysis is worthy of further exploration, and it’s also a problem policy makers eager to solve.Researches of diabetes’economic burden in China started late. Many scholars were conducting such kind of researches, but these studies were fragmented with variety of study areas, contents, methods and findings. The imperfect of national health information systems and missing or unreleased of related basic data lead to so less national researches of diabetes’economic burden. Literatures on forecasting research of diabetes’ economic burden are few. What’s worth, rough prediction methods, too early study time and poor timeliness made previous researches difficult to contribute in policy making. Therefore, it’s urgent to draw on the experience and research methods of developed countries and to explore the theoretical basis and research ideas of diabetes’economic burden prediction, to build prediction model in China, to apply in health research and planning practice, and to service for health policy making and resource allocation in China. This paper will systematically summarize related theories and models on diabetes’ economic burden prediction, in order to build forecast model of this study, to predict diabetes’ economic burden in future years, to explore factors driving the growth of diabetes’ economic burden, to provide a scientific basis for health planning and resources allocation, to achieve "needs-based" health services planning and provision and to ensure the investment of resources match future requirements.Based on the above, this paper proposes the following research questions:1) How many prediction models for diabetes’ economic burden are there in domestic and foreign countries? What’s the theoretical basis, advantages and disadvantages of each model?2) How to build prediction models for diabetes’economic burden? What are available data?3) How is the historical trend of diabetes’economic burden in China? Which are motivating factors of its growth?4) What the future development of diabetes’economic burden will be? Which are driving factors?ObjectivesThe overall objective of this thesis is to explore how to build diabetes’ economic burden forecasting model by researches of theoretical models and empirical analysis, to predict future development of diabetes’ economic burden by analyzing its historical trend and to provide data and scientific evidence for health planning and resources allocation by its motivating factors. Specific objectives including:1) to systematically summarize the related theories and models on diabetes’economic burden prediction in domestic and foreign countries;2) to build forecast models of diabetes’ prevalence and economic burden;3) to analyze the historical trend and its movitating factors of diabetes’economic burden during1997-2009;4) to forecast the future development of diabetes’economic burden and its motivating factors;5) to provide the scientific suggestions for national health planning and resources allocation.MethodsThe strategy of this study of prediction study of economic burden of Diabetes, is to predict the future population, diabetes prevalence and the average of disease economic burden per capita respectively based on historical trend, and then generate the final grossdisease economic burden of diabetes for some year by multiplying the results above together. The prediction study is conducted based on group model by age. The disease economic burdens of diabetes include disease burden, direct economic burden and indirect burden.Firstly, generate group prediction model of diabetes prevalence and individual prediction model of diabetes at the basis of summarial analized the existing prediction model of the forein country.Trend extrapolation and function model are applied to predict diabetes prevalence, while the influencing factors of the disease occurrence rate are analyzed with panel data regression model; secondly, generate the prediction model of disease economic burden per capita, medical service utilization and cost prediction model. The trend extrapolationand annual growth rate method is used for the former one, while sample selection model is applied to the later model. The evaluation and measurement of disease economic burden of diabetes focuse on disease economic burden caused by all the diseases and additional economic burden caused by diabetes, which is based on measurement method of direct economic burden of diabetes recommended by WHO; The disease burden is evaluated by patient numbers and labor-days lost and the economic burden is evaluated by cost of illness approach, while direct economic burden is evaluated by medical fees and indirect economic burden is evaluated by human capital approach. Finally, Logarithmic Mean Divisia Index is applied to decompose impacting factors’contribution to the economic burden growth.The study is based on the data of China Health and Nutrition Survey (CHNS) which was a long term time-series international cooperation program conducted by North Carolina University and China Center for Disease Control. Given various economic development level and dietary pattern, the study covered9provinces including Liaoning, Heilongjiang, Jiangsu, Shandong, Henan, Hubei, Hunan, Guangxi, and Guizhou. The cross-sectional analysis was conducted every2-4years on4400households selected randomly, involving19000subjects. As the diabetes investigation started from1997, therefore, we only use data from investigations in1997,2000,2004,2006and2009. Adults age20and above from the five investigations are included. The records of one subject ranged from1to5, and there are48696records in all. All the diabetic patients involved were diagnosed patients. The demographic data is based on predicted database from US Bureau of the Census. The study describe the demographic characters of subjects, investigated the factors influencing diabetes prevalence and medical fees, analyze the historical trend of disease economic burden of diagnosed diabetes in China from1997to2009, and predicte the future trend of the burden from2010to2020. The indexes involve physical condition, health related factor, social demography economic characteristics, medical service utilization and disease economic burden. Different index variables are analyzed with different statistical analysis methods. Chi-square test and variance analysis are applied to the divergence analysis.Results1. Basic characteristics of research subjectThe subject of this research was the populations who were20+years old. The percent of female was about52%over the period from1997to2009; the percent of the group (20-39ages) decreased about10%, while for the group (45-59ages) and group (60+ages), the percent increased about10%respectively from1997to2009. For the group whose education level was no higher than primary school, its percent decreased about8%, while for the groups who attended junior middle school, senior middle school and the university, the percents of these three groups increased about3%respectively. The percent for the unemployed increased17%, and the retirees raised about5%from1997to2009. There were about33.3%people who owned town registered permanent residence. The percent of the population with medical insurance raised about64%, and household incomes per capital increased from4013yuan to11487yuan. Compared with people without diabetes, the age structure of diabetic patients was more ageing; and the overall education level was lower, the proportion of unemployed and retirees was higher, and the proportion of town census register was higher.The four-week prevalence rate increased by about8%of total population, the prevalence of chronic diseases increased by about9%, obesity rate had increased about5%, the rate of overweight increased by about14%, hypertension rate increased by about7%, excessive physical activity rate fell by nearly15%, smoking and drinking rate decreased about4%, respectively. According to the comparison, for the diabetic population, its four-week prevalence rate, obesity rates, the rates of overweight and high blood pressure were significantly higher than people without diabetes; however, the rate of excessive physical activity, drinking and smoking among diabetic population was lower than people without diabetes.2. Influence factors of diabetes prevalence and medical expensesDiabetes prevalence between different characteristics of grouping people analysis results showed that the diabetes prevalence was different among different groups divided by ages, education, occupation, income and household registration (P<0.05). The diabetes prevalence was higher among those groups (aged60+, lowest education level and highest education level, the retirees and the unemployed, low-income and high-income). The differences on the diabetes prevalence among different groups which were divided by different characteristics, such as different body mass index, blood pressure, physical activity, smoking, drinking and so on, were statistically significant (P<0.05). The diabetes prevalence of overweight group was2times more than the normal and lighter groups, the obese group was four times more than the normal and lighter groups, hypertension group was about3times of the group without hypertension, and it was3times for group without excessive physical compared with group with excessive physical activity. Diabetes prevalence of multi-factor analysis showed that age, obesity, high blood pressure, excessive physical activity and other factors had significant influence on the diabetes prevalence (P<0.05). Results of analysis on the medical expenses per case among groups divided by different characteristics showed that there were statistically differences between different groups which were divided by gender, age, education, occupation, income, household registration kinds of medical insurance. Male was about1.37times than that of women, and group (aged60+) was about1.53times than that of group (aged20-39). Groups (high education level, the retirees and the unemployed) had higher medical expenses per case than other corresponding groups. However there were no significant differences between different groups which divided by gender, age, education, occupation, income, household registration and kinds of medical insurance on the medical expenses per case (P>0.05).Medical expenses of multi-factor analysis results showed that age, marital status, educational level, occupation, medical insurance, family annual per capita income, household registration and whether in hospital onthe influence of medical expenses were statistically significant (P<0.05).3. Historical trend of disease and economic burden for diagnosed diabetes in China From1997to2009, the diagnosed diabetes prevalence rate increased from0.95%to2.24%among people aged20+years old, and that rose0.07%,1.11%and4.15%respectively in the group aged20-39,40-59and60+. The diagnosed diabetes patients rose from7.7017million to21.5959million, which included0.3058million (aged20-39),5.6993million (aged40-59) and7.8691million (aged60+). There were2.6530million increased diagnosed diabetes patients resulted from demographic changes, and8.0741million increased patients resulted from the rise of prevalence rate. There were3.1471million increased patients under the unit effect of demographic changes and the rise of prevalence rate. The total days due to illness among diagnosed diabetes patients in China rose from97.08million days to97.08million days, and the total growth rate was283.94%, while the growth rate caused by the increase number of patients was199.94%, and the growth rate of28.94%caused by the increased days due to illness per capita, and the growth rate was55.06%which was affected by those two. Extra days due to illness rose from41.91million days to41.91million days. The total growth rate was457.14%, and the growth rate of179.20%was caused by the increase of the number of patients and the growth rate of93.38%was caused by the increased extra days due to illness per capita. A combination caused the growth rate of184.56%. Total medical costs of diagnosed diabetes patients in China increased from24.719billion yuan to97.893billion yuan. The total growth rate was296.02%. The growth rate of190.79%was resulted from the increased number of patients, and the medical expense per capita led to the growth rate of36.64%. The growth rate of68.59%was made by the combine effect of those two aspects. Additional medical costs increased from19.719billion yuan to68.295billion yuan, the total growth rate of246.34%. The growth rate of180.73%caused by the increase number of patients and the increase of additional medical expense per capita led to the growth rate of25.16%, and the combination of those caused the growth rate of40.45%. The total indirect economic burden to diagnosed diabetes patients in China increased from2.007billion yuan to26.117billion yuan, and the total growth rate was1201.53%.The growth rate of199.47%was caused by the increase number of patients and the increased indirect economic burden per capita led to the growth rate of340.49%, and a661.57%growth rate resulted from the combination cause; Additional indirect economic burden was from866million yuan to16.363billion yuan, and the total growth rate was1498.90%. The growth rate of179.26%was caused by the increase of the number of patients, and the extra indirect economic burden increase per capita led to the growth rate of563.73%.The growth rate of1045.91%was built by a combination effort.4. Projection of disease and economic burden for diagnosed diabetes in China From2010to2020, the diagnosed diabetes morbidity of people over20will rise from2.313%to4.664%, of which the aged20-39group, the aged40-59group and the aged60+group will rise0.018percentages,1.673percentages and5.815percentages respectively. The number of diagnosed diabetes patients will rise from22.8555 million to50.1356million, of which the20-39group will decrease33.6thousand, the40-59group will increase8.0266million and the60+group will increase19.2841million, and the total growth rate of diagnosed diabetes patients will be119.35%, of which caused by demographic change is25.86%, morbidity rise is71.97%and the combined effect of both is21.70%. The total number of absence days from work of diabetes patients will rise from403.63million to1084.51million and the total growth rate will be168.69%, of which caused by the rise of patients number is128.63%, the rise of absence day from work per capita is19.74%, and the combined effect of both is22.12%. The total number of extra absence day from work will rise from256.27million to773.79million, and the total growth rate will be201.95%, of which caused by the rise of patients number is123.51%, the rise of extra absence day from work per capita is33.96%and the combined effect of both is44.48%. The total medical expense of diagnosed diabetes patient is predicted to rise from109.415billion to402.053billion, and the growth rate will be267.46%, of which caused by the rise of patients number is125.11%, the rise of medical expense per capita is63.47%and the combined effect of both is78.88%. The total extra medical expense will rise from75.707billion to254.165billion, and the total growth rate will be235.72%, of which caused by the rise of number of diagnosed patients is117.39%, the rise of medical expense per capita is55.34%and the combined effect of both is62.99%. The total indirect financial burden of diagnosed diabetes patients will rise from30.544billion to169.127billion, and the total growth rate will be453.71%, of which caused by the rise of number of patients is128.63%, the rise of indirect financial burden per capita is143.06%and the combined effect of both is182.02%. The total extra financial burden will rise from19.393billion to120.672billion, and the total growth rate will be522.25%, of which caused by the rise of number of patients is123.45%, the rise of extra financial burden per capita is176.02%and combined effect of both is222.78%.Conclusions and RecommendationsForecasting studies of diabetes’economic burden in China has a crucial importance for health planning and resources allocation to meet the demand of future. The key point and difficulty of predicting studies is the build of prediction model and the acquisition of data, which is the main reason for so few researches in domestic. Based on systematically summarize related theories and models on diabetes’ economic burden prediction in domestic and foreign countries, this thesis carries out quantitative predictions of China’s future economic burden of diabetes. China is in the economic and social transition that diabetes’prevalence is rapidly growing. The higher level of province’s economic development, the higher prevalence of diabetes, prevalence in urban area is higher than the one in rural area, the same in one area. Prevalence of diabetes in different populations is different, and there is higher prevalence in the old, low-educated, informal workers, the low-income and other vulnerable populations. Medical expenses in different populations also different, the old and informal workers and other vulnerable populations also have higher prevalence. However, the cost differences among different populations are small, which indicates heavy burden for all diabetes patients. Compared with non-diabetes patients, the utilization rates of healthcare and costs per time of diabetes patients are higher, which lead to higher average annual per capita cost and larger individual economic burden and family economic burden. The prevalence of diabetes has brought heavy burden to population health and social economy, even heavier in the future. As the rapid growth of factors, healthcare expenditure, social development and aging population, the prevalence rate, patients and economic burden of diabetes in China will increasing dramatically. The increase of prevalence plays a more important role to increased patients than population changes. The contribution of diabetes patients’ increase to disease economic burden increase is greater than the one of increase cost of per patient. While, the contribution of per patient’s cost also can’t be ignored. Only with effective measures, can we avoid the "blowout" phenomenon of diabetes’ economic burdea.This study has the following policy implications:1) In order to make better predictions of future health problems, relevant department should strengthen the collection of data and improve electronic information system of all levels health institutions.2) According to the trend and motivating factors of national diabetes’ economic burden, we can carry out health planning and resources allocations to meet the demand of future.3) In order to control and delay the economic burden of diabetes, it’s necessary to take effective health economic policy and preventive interventions, such as diet control, reduce obesity and high blood pressure, etc., to strengthen diabetes prevention interventions and disease management for reducing diabetes’ economic burden from the source.4) For those suffer heavy economic burden, specific health policy and assistance should be adopted, such as covering diabetes outpatient services into medical insurance payments, expanding medial insurance reimbursement for diabetes patients. So, it’s possible to reduce economic burden of diabetes patients and improve the life quality of them.Innovations and LimitationsThe innovations including:1) There has not been systematically study of burden and economic burden of diabetes. Different studies using different prediction models got quite different results. Based on literature review, this thesis systematically summarizes related theories and models on diabetes’economic burden prediction in domestic and foreign countries, compare and evaluate their strengths and weaknesses, and come up with suggestions for improvements and future development direction by ideas and methods of mathematical statistics and econometrics.2) There is a lack of domestic economic burden prediction studies of dabetes. Based on study of theoretical models, this paper carries out prediction study of national diabetes’future economic burden, explores answers to the questions we may encounter on the progress, in order to not only to provide data and scientific evidence for health planning and resources allocation but also open a new frontier for forecasting studies of other health problems.3) Panel data model and sample selection model are used to analyze the factors of diabetes and healthcare expenditure. Advantages of both models make research findings more reliable, at the same time, it’s helpful to predict the future development of diabetes’economic burden by analyzing the trend of influencing factors. By calculation method reference of diabetes’direct economic burden, recommended by WHO, and estimating patients’additional economic burden of diabetes by comparing analysis, it’s possible to evaluate diabetes’economic burden and show heavier economic burden of diabetes patients than non-diabetes patients, even though it’s impossible to separate health losses and economic losses of patients. The limitations and prospects including:1) In this thesis, only trend extrapolation is used, the prevalence of diabetes prediction using mathematical functions, per capital economic burden prediction using the annual growth rate, both of which attribute the role of all factors to time variable and make factors of prevalence and healthcare expenditure two separate parts. This is the limitation of methodology in this study. The subsequent research can contribute to macroeconomic panel data regression model, microscopic simulation model and long-term costs model of disease.2) We only utilized a few years’data, mainly because historical data don’t available in a short time by survey and domestic related data are not published, which may affect the accuracy of long-time prediction results to a certain extent. It’s better to track and collect long-term relevant data in the future, and as time goes on list the latest information into time series data to re-fitting the model. In addition, there are many complex factors influencing the prevalence and healthcare expenditure. The future changes of socio-economic and policy environment will reduce the predicting effect of original model, so dynamic scenario analysis can be adopted in the further study.3) Its still in initial stage for diabetes’ economic burden prediction research with inadequate theoretical models to be developed in the further study. The prediction of diabetes’economic burden is prospective study, predictions of various methods are only scientific estimations, every model is simplify and abstraction of development process. So they only provide information from a certain angle, and they can’t fully reveal its development and changes. Thus, many predictive models can be constructed in the future, by evaluating to determine the optimal model.更多还原