【作者】 周琦；

【导师】 王卫平；

【作者基本信息】 复旦大学 ， 儿科学， 2014， 博士

【摘要】 前言新生儿重症监护病房(neonatal intensive careunit, NICU)医院感染(nosocomial infection, NI)发生率高,是危重新生儿和早产儿发病及死亡的重要原因。随着围生医学的发展和新生儿重症监护救治水平的提高,早产儿成活率明显提高,但成活患儿的疾病发生率并未降低。NICU内极低／超低出生体重(very low birth weight infant, VLBW和extremely low birth weight infant, ELBW),因免疫功能极不成熟,多数存在基础疾病,救治过程中往往需要接受侵入性有创诊疗措施,是NICU发生医院感染的高危人群。而装置相关感染(device-associated infection, DAI)作为一种特殊的医院内感染也引起越来越多国内外新生儿医护工作者的高度关注,其中呼吸机相关肺炎(ventilator associated pneumonia, VAP)和中心静脉置管相关血流感染(Central-line associated bloodstream infection, CLABSI)是NICU内装置相关感染的主要类型,也是引起VLBW和ELBW死亡、发生并发症以及神经发育落后的重要原因。有效防治医院感染尤其是装置相关感染可降低危重新生儿及早产儿病死率,减少住院时间及医疗费用,改善患儿近期及远期预后。VAP指因非肺部感染性疾病经气管插管行机械通气48 h后,或因感染行机械通气48h后肺部出现新的感染,为医院获得性肺部感染的主要类型。由于新生儿免疫系统尚未发育成熟,而机械通气可破坏其会厌部正常屏障功能,增加将口咽部微生物直接定植于下呼吸道的风险,此外对呼吸机管道、湿化器和复苏囊等消毒不严格,医务人员操作前后未严格执行手卫生制度、滥用抗生素等均可导致VAP的发生。近十余年来国内医院广泛建立NICU,使得越来越多的早产儿得到了救治及存活的机会,然而很多早产儿需要呼吸机辅助通气,由于早产儿免疫功能低下,感染发生率高,VAP成为NICU面临的重要挑战。VAP治疗棘手,严重制约了机械通气的效果,影响患儿预后且增加医疗费用。CLABSI指放置血管内导管48小时后,或者拔除血管内导管48小时内出现菌血症或真菌血症,研究表明导管相关血流感染是院内晚发型败血症的主要类型3。中心静脉置管穿刺创伤小、保留时间长,并可避免因多次穿刺或输高渗液对血管的损害,同时在很大程度上减轻了患儿反复静脉穿刺的痛苦,是危重新生儿及极低／超低出生体重儿长期使用肠外营养及高渗补液的重要途径。早在1973年,国外即报道4将经外周置入中心静脉导管(peripherally inserted central catheters, PICC)作为可靠的静脉通路为危重新生儿提供全肠外营养,随后PICC成为发达国家NICU的常规治疗手段。我国仅在近十余年开始将中心静脉置管技术广泛应用于NICU。导管相关血流感染是应用中心静脉置管最严重的并发症之一,可延长患儿住院天数,增加患儿病死率及医疗负担,严重影响患儿的治疗和预后。因此,如何减少或避免中心静脉置管相关血流感染,以提高NICU患儿抢救成立率及改善患儿预后,是目前临床亟待解决的重要问题。由于危重新生儿尤其是极低出生体重儿免疫力低下,感染临床表现具有隐匿性、非特异性的特点,患儿各器官功能发育未成熟,发生严重感染后,即使在使用有效抗生素、积极对症支持治疗下,患儿的病情依然进展迅速,可在数小时进展为脓毒血症、多器官功能衰竭,最终导致死亡。因此早期识别并干预新生儿严重感染对降低新生儿病死率显得尤为重要。然而目前临床上尚缺乏早期、快速、有效的感染辅助诊断及预警方法,各种标本培养阳性率低,且需要至少48小时发现细菌生长。其它生物学指标如C-反应蛋白、血清降钙素原(procalcitonin, PCT)、细胞因子、Toll样受体等在新生儿感染的诊断中有一定价值,但其敏感性或特异性仍较低,且这些指标在疾病发生和发展过程中呈动态改变,需要对患儿采血进行连续动态检测,其临床应用价值受到限制。NICU患儿的生理信息数据处于动态变化中,床旁医疗监护设备可对这些数据进行连续监测并提供大量的信息。某些生理信息资料的细微改变可能对即将发生的严重临床事件具有预测价值。然而目前大量的监护仪数据和信息未经详细分析处理既已丢失。NICU尚缺乏可最大限度利用这些信息实时帮助临床诊疗和进行回顾性临床分析研究的设施、工具和技术,大量的数据和信息未发挥其利用潜能。感染早期由于促炎细胞因子的释放,可影响新生儿未成熟的自主神经系统及内环境稳态,进而影响循环及呼吸系统,导致血压、心率及呼吸频率等生理参数的特征性改变。早期监测这些生理参数的改变,并对大量生理参数进行深入研究,有望为临床对败血症等严重感染提供早期预警。目前发达国家提倡应用基于循证实践的病房质量改进措施,使用美国疾病预防控制中心(centers for disease control and prevention, CDC)制定的发病密度(incidence density)的监测方法对NICU内VAP及CLABSI的发生进行持续监测,提倡应用“集束化”方法将目前已证实有效的一系列操作、治疗、护理等措施集合在一起对医院内感染进行防治5,并开始探索应用先进的质量管理方法对持续改进医疗质量的效果。部分NICU甚至做到了对DAI的“零容忍”。同时发达国家开始探索将信息技术与临床专业知识结合,研究如何使用生理“大数据”早期预警新生儿感染的发生,为新生儿重症监护提供崭新的临床决策及高质量循证支持。与发达国家相比,我国机械通气及经外周中心静脉置管在新生儿的应用起步较晚,一些医护人员的感染防治意识仍然相对薄弱,目前VAP及CLABSI已是国内NICU面临的重要问题,然而相关报道较少。已报道的DAI发生率仍然使用传统的发病率监测,对DAI定义尚未完善,因此不能与发达国家进行横向比较,且国内尚缺乏DAI相关的临床干预研究及防治指南,多数NICU仍然采用的是经验性、零散的感染防治措施。同时,如何在信息化时代发挥学科交叉的优势,充分挖掘利用NICU内监护仪采集的生理信息“大数据”潜能,探索新生儿严重感染早期预警方法,帮助临床医生早期发现及干预NICU医院内感染,进而降低早产儿感染相关的并发症及死亡率也是摆在我们面前的崭新课题。本研究使用CDC建议的标准监测方法对VAP及CLABSI的发生进行了前瞻性连续监测,并在我院NICU内实施装置相关感染综合干预措施；在国内首次搭建床旁监护仪生理数据实时采集存储平台,对感染及非感染患儿的生理数据流进行初步的数据挖掘及时程分析。对干预措施实施前后医院内感染的持续监测旨在了解并获得本院可靠的VAP及CLABSI发生的相关流行病学数据,观察干预措施的效果,探索有效的VAP及CLABSI防治措施,为临床对NICU院内感染的干预提供指导,也为将来制定适合国内NICU的VAP及CLABSI防治指南提供依据。同时,本研究通过搭建NICU监护仪生理数据平台,对患儿生理数据进行采集和数据挖掘,旨在探索预警NICU医院感染发生的生理信息模式,希望通过此研究,在国内NICU推广信息医疗的理念,对通过捕捉并利用患儿呈现的细微生理信息数据变化来预警即将发生的临床事件进行初步探索,为下一步开展更为深入的患儿生理信息大数据挖掘提供平台及基础。本研究经复旦大学附属儿科医院伦理委员会批准。第一部分综合措施防治新生儿重症监护病房呼吸机相关肺炎的研究背景随着中国医疗水平的提高及经济的快速发展,国内医院纷纷建立NICU,使得越来越多的早产儿得到了救治及存活的机会,然而大部分早产儿需要呼吸机辅助通气,新生儿免疫功能发育未成熟,感染发生率高,因此VAP成为NICU面临的重要挑战。VAP的发生与危重新生儿发病及死亡密切相关,延长患儿住院时间,增加患儿住院费用,在发展中国家尤为显著。发达国家NICU建立较早,已对VAP防治进行了大量研究,并形成了一定规模的监测体系及科学管理方法。而发展中国家尚缺乏对NICU内VAP监测及干预措施效果的相关研究报道。目的1.监测我院NICU内VAP发生情况,了解VAP的病原菌分布及耐药性变迁；2.探讨NICU不同时段采用“集束化”综合干预措施对VAP发生率的影响,为临床提供有效的防治策略。方法 本研究采取前瞻性前后对照研究。以2008年6月复旦大学附属儿科医院搬迁至新院为分时段标志,第一阶段为搬迁前1年,2006年2月1日至2007年1月31日,旧院NICU环境(干预前)；第二阶段为搬迁后第一年,2008年8月1日至2009年7月31日,新院NICU环境(部分干预)；第三阶段为搬迁后第二年,2010年1月1日至2010年12月31日,新院NICU环境(系统干预)。研究对象为气管插管机械通气≥48 h,并在NICU住院≥5d的患儿,收集相关临床及实验室检查资料。对三个阶段VAP的发生及其病原菌分布和耐药性变迁进行监测。应用统计软件STATA10.0进行统计分析。结果1.研究期间符合纳入标准的机械通气新生儿共491例。机械通气新生儿占住院新生儿比例第一阶段为5.0%,第二阶段为6.8%,第三阶段为7.8%。机械通气最主要的原因为早产儿呼吸窘迫综合征,占47.9%。2.491例机械通气新生儿中,92例发生VAP,机械通气天数为3366天,研究期间VAP总的发生率为27.3/1000呼吸机使用日。第一阶段VAP发生率为48.8/1000呼吸机使用日；第二阶段VAP发生率为25.7/1000呼吸机使用日；第三阶段VAP发生率为18.5/1000呼吸机使用日。三个阶段VAP发生率逐年下降,差异有统计学意义(P<0.001)。纳入的机械通气患儿的死亡率分别为第一阶段14.0%,第二阶段2.9%,第三阶段2.7%。三个阶段机械通气患儿的死亡率逐年下降,差异有统计学意义(P<0.001)。3.92例VAP患儿下呼吸道痰液标本共分离出病原菌66株,混合感染3例。病原菌以革兰阴性杆菌感染为主,共计63株,占95.5%(63/66)。鲍曼不动杆菌是VAP最主要的致病菌,共计43株,占65.2%(43/66)；其次是肺炎克雷伯菌10株(10/66),占15.2%。4.3个阶段相比,第一阶段致VAP的鲍曼不动杆菌对第三代头孢菌素全部耐药,对碳青霉烯类抗生素敏感率为57.9%,有8株多重耐药菌株；采取综合感染防治措施后第二阶段分离到的病原菌对第三代头孢菌素敏感率上升至75.0%,与第一阶段比较,差异有统计学意义(P=0.000),且无耐碳青霉烯类菌株及多重耐药菌株；与第二阶段相比,第三阶段病原菌对头孢菌素及碳青霉烯类药物的敏感率分别下降至5.0%(P=0.001)和20.0%(P=0.01),出现10株多重耐药菌株。结论1.采用“集束化”综合感染防治措施能有效且持续地降低VAP的发生率。2.与发达国家相比,我院NICU内VAP发生率仍较高,有待进一步研究探讨防治策略。3.VAP病原菌以革兰阴性菌为主,鲍曼不动杆菌是VAP最主要的致病菌。4.VAP病原菌种类逐年变迁。定期监测和分析NICU患儿VAP病原菌分布及其耐药性变迁,对指导临床合理使用抗菌药物及控制VAP发生具有重要意义。3.92例VAP患儿下呼吸道痰液标本共分离出病原菌66株,混合感染3例。病原菌以革兰阴性杆菌感染为主,共计63株,占95.5%(63/66)。鲍曼不动杆菌是VAP最主要的致病菌,共计43株,占65.2%(43/66)；其次是肺炎克雷伯菌10株(10/66),占15.2%。4.3个阶段相比,第一阶段致VAP的鲍曼不动杆菌对第三代头孢菌素全部耐药,对碳青霉烯类抗生素敏感率为57.9%,有8株多重耐药菌株；采取综合感染防治措施后第二阶段分离到的病原菌对第三代头孢菌素敏感率上升至75.0%,与第一阶段比较,差异有统计学意义(P=0.000),且无耐碳青霉烯类菌株及多重耐药菌株；与第二阶段相比,第三阶段病原菌对头孢菌素及碳青霉烯类药物的敏感率分别下降至5.0%(P=0.001)和20.0%(P=0.01),出现10株多重耐药菌株。结论1.采用“集束化”综合感染防治措施能有效且持续地降低VAP的发生率。2.与发达国家相比,我院NICU内VAP发生率仍较高,有待进一步研究探讨防治策略。3.VAP病原菌以革兰阴性菌为主,鲍曼不动杆菌是VAP最主要的致病菌。4.VAP病原菌种类逐年变迁。定期监测和分析NICU患儿VAP病原菌分布及其耐药性变迁,对指导临床合理使用抗菌药物及控制VAP发生具有重要意义。第二部分 综合措施防治新生儿重症监护病房中心静脉置管相关感染的研究背景近十余年来,中心静脉置管在国内NICU得到迅速推广应用,成为NICU重要的救治手段。然而由于NICU内置管对象多为危重新生儿和极低/超低出生体重儿,其机体各组织器官功能尚未成熟,免疫功能低下,且PICC导管留置时间较长,极易发生中心静脉置管相关感染(CLABSI)。CLASBI严重影响患儿的治疗效果及预后。国外发达国家已有大量研究对CLABSI的有效防治措施进行探索,提出“集束化”干预的概念,并开始采用质量管理方法对CLABSI进行防治,部分NICU内CLABSI发生率已降至0。国内在CLABSI防治领域仍然处于零散的经验防治阶段,对CLABSI的监测及防治措施的效果也不甚明确。目的1.监测我院NICU内CLASBI发生情况,了解CLABSI的病原菌分布；2.研究“集束化”综合防治措施对NICU内CLABSI的干预效果,为临床防治CLABSI发生提供理论及实践依据。方法本研究为前瞻性前后对照研究,对三个不同时期2008.1.1-2008.12.31(第一阶段：干预前阶段)、2009.1.1-2009.12.31(第二阶段：干预措施实施阶段)及2010.1.1-2010.12.31(第三阶段:干预后随访阶段)CLABSI发生情况进行分析。研究对象为收入复旦大学附属儿科医院NICU,使用中心静脉置管≥48小时,并在新生儿病房住院时间≥5天患儿,收集相关临床及实验室检查资料。比较采取干预措施前后三个阶段CLABSI发生情况及其病原菌分布。应用统计软件STATA 10.0进行统计分析。结果1.研究期间符合纳入标准的中心静脉置管新生儿共171例,共计住院10399天。其中第一阶段纳入患儿29例,第二阶段纳入患儿51例,第三阶段纳入患儿91例。三个阶段纳入患儿的临床一般资料无显著差异。2.171例中心静脉置管新生儿中,29例发生CLABSI, CLABSI发生率为17.0%。其中干预前第一阶段CLABSI发生率为16.7/1,000置管日,干预措施实施的第二阶段CLABSI发生率为7.6/1,000置管日,随访监测的第三阶段CLASBI发生率为5.2/1,000置管日。三个阶段CLABSI发生率逐年下降,差异有统计学意义(P<0.001)。3.29例CLABSI患儿血液标本共分离出病原菌22株,混合感染1例。总体上,病原菌以革兰阴性杆菌感染为主,共计12株,占54.5%(12/22),最主要的革兰阴性菌为鲍曼不动杆菌及肺炎克雷伯菌,各占27.3%(6/22)及18.1%(4/22)。凝固酶阴性葡萄球菌是CLABSI最主要的致病菌,共计7株,占31.8%(7/22)。三个阶段比较,可见革兰阳性菌所占比例有上升趋势(第一阶段：16.7%；第二阶段：40%；第三阶段：36.3%)。结论1.“集束化”CLABSI综合感染防治措施能有效且持续地降低NICU内CLABSI的发生率。2.与发达国家相比,我院NICU内CLABSI发生率仍较高,有待进一步研究探讨防治策略。3. CLABSI病原菌以革兰阴性菌为主；凝固酶阴性葡萄球菌是CLABSI最主要的致病菌。4. CLASBI病原菌种类逐年变迁,三个阶段比较,革兰阳性菌所占比例有上升趋势。定期监测和分析NICU患儿CLABS1病原菌分布及变迁,对指导临床合理使用抗菌药物及控制CLABS1发生具有重要意义。致病菌。4. CLASBI病原菌种类逐年变迁,三个阶段比较,革兰阳性菌所占比例有上升趋势。定期监测和分析NICU患儿CLABS1病原菌分布及变迁,对指导临床合理使用抗菌药物及控制CLABS1发生具有重要意义。第三部分 生理数据分析在新生儿感染监测中的应用研究背景 新生儿感染是引起新生儿发病及死亡的重要原因。新生儿严重感染早期释放的促炎症细胞因子可通过自主神经系统、神经-体液免疫反应及细胞水平的复杂交互作用影响机体,引起心率、呼吸、血压等生理参数变化。如能将NICU监护仪连续采集的心率、呼吸频率、氧饱和度、血压等大量的生理数据信息进行存储,通过数据挖掘及分析发现患儿出现感染临床表现前的生理数据信息变化规律,有望在患儿发生严重感染的早期提出预警以指导临床早期诊治,改善患儿预后。目前NICU尚缺乏可最大限度利用监护仪采集的生理大数据信息实时帮助临床诊疗和进行回顾性临床分析研究的设施、工具和技术,大量的数据和信息未发挥其利用潜能。近年来研究显示：将采集的生理数据、信息与专业知识进行迅速准确的整合,可为临床决策提供高质量循证支持,帮助临床早期预测某些感兴趣的事件发生或是评估疾病的预后。目的 1.在复旦大学附属儿科医院NICU内建立可实时采集和保存高保真生理数据流的系统环境,实现对监护仪采集的患儿生理数据信息进行长期存储及回顾性利用；2.分析心率变异性(Heart rate variety, HRV)及呼吸频率变异性(Respiratory rate variety, RRV)在新生儿感染尤其是新生儿败血症早期的变化,探讨其在NICUI临床应用价值。方法 借鉴加拿大多伦多儿童医院及安省理工大学相关研究经验,并与对方合作,在复旦大学附属儿科医院NICU内建立可采集和存储临床患儿监护仪高保真生理数据流的Artemis系统环境。自2013年4月1日至2013年9月30日通过搭建的Artemis平台前瞻性纳入病例,并开始正式采集监护仪数据。使用合作研究者McGregor教授提供专利的STDMno框架及SAS统计软件完成对纳入患儿HRV及RRV的时间序列分析及数据挖掘。将数据挖掘的结果与临床事件相对应,比较败血症患儿同非败血症患儿、败血症患儿本身发生感染前后HRV及RRV的变化,初步探讨新生儿感染发生前后HRV及RRV的变化规律。结果1.成功搭建Artemis系统环境：在复旦大学附属儿科医院新生儿重症监护病房建立了Artemis高保真生理数据采集环境,将21台床旁监护仪采集的生理数据(血氧饱和度、呼吸频率、心率、血压)以高保真、高频率数据流的方式通过IBMstreams分别编排分装、实时传输并存储至DB2数据库中。实现生理数据流永久性的本地存储及回顾利用。2.研究共纳入患儿80人,共37人发生38例次感染,包括29例次晚发型败血症(8例确诊败血症及21例临床败血症)、8例尿路感染及1例脑膜炎。定性分析结果显示：确诊败血症感染患儿HRV降低,RRV正常。3. 43例患儿生理数据流格式严格符合时间序列分析及数据挖掘标准,包括6位败血症患儿(3例次血培养阳性败血症,3例次临床败血症)及37位未感染患儿。败血症患儿感染前48小时(30.42±6.06min/hour VS 33.6±6.02min/hour, P=0.18)、感染前96小时(30.2±6.02min/hour VS 33.6±6.02min/hour, P=0.17)、及全病程(29.5±4.72min/hour VS 33.6±6.02min/hour, P=0.08)平均每小时正常HRV时程与非败血症患儿全病程相比均降低,但无统计学差异。4.将败血症患儿自身感染前后不同时间窗的心率变异性进行配对比较1)败血症临床诊断前平均每小时正常HRV时程为29.24分钟,败血症治疗48小时后平均每小时正常HRV时程为33.62分钟,败血症临床诊断前较败血症治疗48小时后平均每小时正常HRV时程显著降低(p=0.003),降低的时程为4.38分钟(95%CI[2.1,6.7])；2)败血症临床诊断前72小时内每小时正常HRV时程较败血症治疗72小时后显著降低(p=0.04),降低的时程为3.83分钟(95%CI[1.4,5.9])；3)败血症临床诊断前72小时内每小时正常HRV时程较败血症治疗1周后显著降低(p=0.006),降低的时程为3.52分钟(95%CI[1.3,5.8])。结论1.基于国内监护仪型号及接口可实现对床旁医疗监护设备采集到的患儿高频生理数据进行高保真储存并进行多维处理分析；2. Artemis高保真生理数据采集平台的稳定性能尚有待优化和提高；3.初步研究结果显示新生儿败血症前期心率变异性降低；4.新生儿感染前期HRV降低可能存在特定的时间窗,需要将来进行大样本队列研究进一步证实。更多还原

【Abstract】 IntroductionNosocomial infection (NI) is a significant complication for critical ill patients treated in the neonatal intensive care unit (NICU) and results in considerable morbidity and mortality. With advance in perinatal medicine and neonatal intensive care, the survival rate of preterm infants is rising obviously. However, the morbidity of preterm infants has not been decreased with the increased survival rate. The infection rate among survivors is still high, especially for very low birth weight (VLBW) and extremely low birth weight infants (ELBW). Neonates hospitalized in NICU have greater likelihood of a critical illness, and invasive monitoring and procedures make them more vulnerable to acquisition of health care-associated infections. The device-associated infection (DAI), especially ventilator associated pneumonia (VAP) and Central-line associated bloodstream infection (CLABSI), are raising more and more attention both from international and domestic neonatal healthcare workers because of worsening neonatal outcomes and associated neurodevelopmental impairment among survivors. Reducing DAI may short length of NICU stay, cut health care costs, decrease the morbidity and mortality among neonates and improve their short- term and long-term prognosis.Ventilator-associated pneumonia (VAP), defined as pneumonia that develops at least 48 hours after introduction of mechanical ventilation (MV) is one of the most common health care-associated infections in neonates. Given the immaturity of neonate’s immune system, MV and other invasive treatment measures are very likely to increase the risk of oropharyngeal or tracheobronchial colonization with pathogenic bacteria, which adhere to the epithelial cells of the respiratory tract. VAP occurs when bacterial, fungal, or viral pathogens enter the normally sterile lower respiratory tract and lung parenchyma. With the rapid development in China, increasing number of neonatal intensive care units (NICUs) are being established across the country. Consequently, increasing numbers of very premature babies are surviving. Because many of them require MV, VAP has become a major challenge. VAP has a large influence on neonatal survival, morbidity, hospital costs, and duration of NICU stay.Central line-associated bloodstream infections (CLABSI), defined as sepsis that develops after 48 hours of central venous catheterization, are a significant source of sepsis in NICU. Central venous catheters or central lines (CL) have been routinely used in NICUs in developed countries to provide long-term venous access for delivering parenteral nutrition (TPN) and medication for over three decades. By contrast, indwelling CL has been only introduced as routine practice in Chinese NICUs during the past decade. CLABSI is one of the most serious complications of CL insertion, which result in considerable morbidity and mortality, increase duration of hospital stay and health cost. Therefore, reduction and prevention of CLABSI are critical to increase the survival rate and improve outcome of neonates treated in NICU.NI can progress to sever sepsis, Multiple organ dysfunction syndrome (MODS) and even death very quickly even with advance antibiotics. The clinical symptoms of NI are often subtle and nonspecific. Infants may often appear asymptomatic until the infection is well established. Therefore, early and accurate diagnosis is challenging. Unfortunately, no single biomarker has proven successful in accurately detecting LONS, and most biomarker tests require serial sampling for to allow a trend to be established to be useful for diagnosis. The need to take multiple blood samples also increases the risk of introducing secondary infection in neonates.In the era of sophisticated electronic information and computerized techniques, bedside medical devices are used in NICU to monitor the clinical condition of neonates. The monitors graphically display complex physiological data for physiological data streams Subtle changes in physiological data may be predictive for a future event. However, large physiological data stream are discarded automatically without further analysis. There are few infrastructure, tools and techniques to utilize this data to its fullest potential to support clinical management and retrospectively analysis.Studies have reported that physiological parameters such as heart rate and respiratory rate would fluctuate and change because of proinflammatory cytokines released in the early stage of infection. Continuously monitor, collect and further analyse these physiological streams, by which to detect its abnormal changes early may provide new ways for non-invasive and early warning system for sepsis and other serious infections in NICU.Currently, quality improvement based on evidence-based practice is prevalent in developed countries, the standard and unified infection surveillance method recommended by CDC is used during infection monitoring process in developed countries. Researchers have created "Care bundles" which is a group of interventions that, when executed together, result in better outcomes than when implemented individually. Meanwhile, healthcare workers in developed countries are trying to combine information technology, health informatic, and clinical expertise to explore the "big data" exist in physiological parameters to early detect neonatal nosocomial infection, which could translate the potential information in neonates’ physiological data into high-quality evidence-based clinical decision support.Comparing with developed countries, healthcare-associated infection, especially DAI control awareness among healthcare workers in NICU is relatively weak, the affect of VAP and CLABSI on health care costs is especially significant in the developing world whereas most relative studies have been conducted in developed countries. Reports on the success of VAP and CLABSI intervention strategies are scarce. Most Physicians in Chinese NICU still practice infection control casually and based on experience. Meanwhile, how to take advantage of interdisciplinary in the information era, explore the "big data" exist in physiological parameters to earlier detection of neonatal infection in NICU is also a frontier topic in front of us.In the first two parts of this study, the "Healthcare bundles" were developed to control VAP and CLABSI in our NICU, VAP and CLABSI incidence density were monitored continuously using the standard calculate method recommend by CDC; in the third part of this study we build a real-time physiological data collection and permanent store platform to do preliminary data mining for earlier detection of neonatal infection which was the first in China. There are two primary aims of the first two parts of this study, one aim is to provide reliable epidemiologic data and characterization of VAP and CLABSI among neonatal patients in our hospital, another aim is to evaluate the efficacy of a multi-faceted infection control program ("bundles") in reducing VAP and CLABSI in our NICU. These two investigations can provide important information for future comparative studies and to encourage and promote the use of surveillance and CLABSI preventive strategies in other Chinese hospitals. By the third part of this study, we aimed to investigate whether there have subtle changes exist in physiological data behaviors present prior to the clinical suspicion of neonatal infection, which should provide an exciting opportunity to advance our knowledge of healthcare informatics application and to offer some important insights into healthcare "big data".The Ethics Committee of children’s hospital at Fudan University approved the study protocol.Part Ⅰ:Efficacy of an infection control program in reducing ventilator-associated pneumonia in a Chinese neonatal intensive care unitBackground:With the rapid development of China, increasing numbers of very premature babies are surviving. Ventilator-associated pneumonia (VAP) has become a major challenge. VAP has a large influence on neonatal survival, morbidity, hospital costs, and duration of NICU stay. The affect of VAP is especially significant in the developing world whereas most relative studies have been conducted in developed countries. Measures employed in preventing VAP in developing countries are rarely reported.Objectives:1. To investigate the incidence, the distribution of pathogens and antibiotic sensitivity of VAP in our NICU2. To evaluate the efficacy of a multi-faucet infection control program (bundles) to reduce VAP in a neonatal intensive care unit (NICU) in China.Methods:This is a prospective before-after study. The hospital relocated to a new site from June 2008 and the intervention was initiated in NICU from phase 2. The first phase was from Feb 1,2006 to Jan 1,2007(pre-intervention phase), the second phase was from Aug 1,2008 to July 31,2009 (partial intervention phase) and the third phase was from Jan 1,2010 to Dec 31,2010(systemic intervention phase). Related data were collected for comparing VAP incidence and pathogen distribution during these three phases after a bundle of infection preventive measures were gradually implemented step by step. STATA 10.0 was applied for statistic analysis in this study.Results:1. A total of 491 neonates were enrolled during the study period. MV patients accounted for 5.0%,6.8%,7.8% among general hospitalized neonates in these three phases. The primary reason for use of MV were premature respiratory distress syndrome (47.9%)2. Of 491 ventilated patients,92 (18.7%) developed VAP corresponding to 27.3 per 1,000 ventilator-days. The rate decreased from 48.8 per 1,000 ventilator-days in phase 1 to 25.7 per 1,000 ventilator-days in phase 2 and further diminished to 18.5 per 1,000 ventilator-days in phase 3 (P<0.001). Overall mortality rate of admitted neonates significantly decreased from 14.0% in Phase 1 to 2.9% in Phase 2 and 2.7% in phase 3 (P=0.000).3.66 strains of pathogens were isolated among 92 patients with VAP, Gram-negative bacilli (63 strains,95.5%) were the primary VAP microorganisms in all phases, and the most common organisms isolated for VAP during these three period were Acinetobacter baumannii(43 strians,65.2%)and Klebsiella pneumoniae(10 strains, 15.2%).4. In phase 1, antimicrobial susceptibility testing showed that all Acinetobacter baumannii were resistant to cephalosporin, and antimicrobial susceptibility of carbapenems were 57.9%, and eight multi-drug resistant bacterial strains were isolated. After implemented the infection-control practices during the second phase, 75% Acinetobacter baumannii were sensitive to cephalosporins, the difference is significant compares to phase 1(P=0.000), and there were no multi-drug resistant or carbapenem-resistant strains isolated in phase 2. However, in phase 3, the antimicrobial susceptibility to the third-generation cephalosporin and carbapenems declined to 5%(P=0.001) and 20%(P=0.01) respectively.10 multi-drug-resistant strains were isolated.Conclusions:1. Implementing a multi-faceted infection control program resulted in significant reduction in VAP rate with long-term effects.2. The VAP incidence was still high compared with the developed countries, further studies are required to investigate the infection control strategies.3. Gram-negative bacilli were the primary VAP microorganisms in our NICU and Acinetobacter baumannii (65.2%) was the most frequently isolated microorganism.4. The pathogens spectrum changes in different periods. Regular analysis and monitoring pathogen’s species and the variance of drug resistance in local patients with VAP are important to guide the clinical use of antimicrobial agents and to effectively control the incidence of VAP.Part II:Study on the Effects of "Care Bundles" in Reducing Central Line-associated Bloodstream Infections in a Chinese Neonatal Intensive Care UnitBackground:Over the past decade, central line (CL) has been introduced to the NICU in China and practiced rapidly as a routine in NICU. Most critically ill neonates, including extremely low birth weight (ELBW) and very low birth weight (VLBW) infants, are frequently managed with CL, which placed this immune-compromised population at high risk of CLABSI. "Care bundles" and scientific quality improvement methodology have been used to reduce central-line associated bloodstream infections (CLABSIs) in developed countries. These efforts have significantly reduced CLABSIs and the CLABSI incidence even fallen to zero in some NICUs in developed countries. Few data are available on CLABSIs in Chinese NICUs. CLABSI prevention in Chinese NICU is still based on experience. Measures employed in preventing CLABSI in developing countries are rarely reported.Objectives:1. To investigate the incidence and the distribution of pathogens of CLABSI in our NICU2. To evaluate the impact of a multi-faceted evidence-based practices (bundles) on the occurrence of CLABSI in our NICU, by which to provide effective strategies for clinical practice.Methods:We conducted a prospective before-after intervention study with a 1-year follow-up among patients with central lines at the NICU of the Children’s Hospital of Fudan University, between January 2008 and December 2010. Patients with CL indwelling for>48 hours and who had been hospitalized in NICU for≥5 days were recruited. The study was conducted in three phases which is before (phase 1,1 January 2008 to 31 December 2008), during(phase 2,1 January 2009 to 31 December 2009), and after (phase 3,1 January 2010 to 31 December 2010) intervention. Multi-faceted evidence-based practices were introduced in phase 2 and continued in phase 3 with successive surveillance. CLABSI were prospectively monitored and compared. STATA 10.0 was applied for statistic analysis in this study.Results:1. A total of 171 patients with CL were observed with the bed-days of 10399. There were 29,51, and 91 enrolled newborns in phase 1, phase2, and phase 3, respectively. There were no significant differences in terms of clinical characteristics between these 3 groups.2. Of 171 patients,29 of them developed CLABSI with the CLABSI rate of 17.0%, and the overall CLABSI incidence density decreased significantly from 16.7 per 1,000 CL-days in phase 1 to 7.6 per 1,000 CL-days in phase 2 and to 5.2 per 1,000 CL-days in phase 3 (P<0.001).3.22 strains of pathogens were identified during the study period, one episode with 2 pathogens was identified. Gram-negative bacterium (54.5%) was the predominant pathogen in CLABSI generally, and the most common Gram-negative bacilli isolated for CLABSI during these three Phases were Acinetobacter baumannii (6 strians. 27.3%) and Klebsiella pneumonia (4strains,18.1%). Coagulase-negative staphylococci (7strains,31.8%) was the most frequently isolated microorganism. Gram-positive bacterium caused CLABSI has an increasing trend during 3 phases.Conclusions1. A multi-faceted infection control program is effective in sustaining reduction of CLABSI rate among neonates.2. The CLABSI incidence was still high compared with the developed countries. Further studies are required to investigate the infection control strategies.3. Gram-negative bacterium was the predominant pathogen of CLABSI. Coagulase-negative staphylococci were the most frequently isolated microorganism.4. The pathogens spectrum of CLABSI changes in different periods. Gram-positive bacterium has an increasing trend during 3 phases. Regular analysis and monitoring pathogen’s species of CLABSI and the variance of drug resistance in local NICU are important to guide the clinical rational use of antimicrobial agents and to effectively control the incidence of CLABSI. Coagulase-negative staphylococci were the most frequently isolated microorganism.4. The pathogens spectrum of CLABSI changes in different periods. Gram-positive bacterium has an increasing trend during 3 phases. Regular analysis and monitoring pathogen’s species of CLABSI and the variance of drug resistance in local NICU are important to guide the clinical rational use of antimicrobial agents and to effectively control the incidence of CLABSI.Part Ⅲ:Earlier Detection of Infection using Temporal Analysis of Continuous Physiological Data Streams in a Chinese Neonatal Intensive Care UnitBackground:Infection is an important cause of morbidity and mortality in neonates. Proinflammatory cytokine released in the early stage of serious infection would result in changes of heart rate, respiratory rate, blood pressure and other physiological parameter via automatic immune- automatic nervous system-heart interactions on the cell level. There is a growing body of evidence to prove the potential to utilize physiological data to assist with the prediction of the onset of clinical events. If the high volume of physiological parameter data from bedside monitor devices could be stored and analysis, it might help physician to find the subtle changes exist in physiological data, which is predictive for neonatal infection. As such the infected neonates can be treated earlier and therefore have better prognosis. There are few infrastructure, tools and techniques to utilize this data streams to its fullest potential to support clinical management and retrospectively analysis. Big volumes of physiological data and information was wasted and not be fully used. The rapid and accurate integration of data from numerous medical systems is essential for quality, evidence-based decision making within the NICU and to help physicians to detect interested clinical events earlier.Objectives:1. To establish an environment which can collect and store the multi-dimensional and high-fidelity physiological data form bedside monitors, by which to realize the long term storage and retrospective analysis of the physiological data.2. To test whether heart rate variability (HRV) and respiratory rate variability (RRV) behaviors present prior to the clinical suspicion of infection especially the neonatal sepsis, by which to explore the application value of HRV and RRV in the infection early detection system.Methods:Cooperate with the research team in The Hospital for Sick Children in Toronto and University of Ontario Institute of Technology, Canada, to set up an Artemis system for our local real-time physiological data collection and storage. We collected physiological data with this local Artemis platform prospectively from April 1st 2013 to September 30th 2013. Using the STDMnO frame provided by the cooperator, Carolyn McGregor, and SAS software together, to do the time series analysis for the HRV and RRV for the enrolled patients. Match the Clinical event with the data analysis result, comparing HRV and RRV between sepsis and non-sepsis patients, and between before and after sepsis treatment for sepsis patients to investigate the changes of HRV and RRV before and after neonatal infection.Results:1. Establishment of local Artemis platform at Children’s Hospital of Fudan University successfully. Given the environment of our NICU, the platform may collect physiological data (SPO2, respiratory rate, heart rate and blood pressure) from 21 bedside monitors synchronously. The data are processed by IBM streams and transfer into the DB2. This platform realizes the physiological data’s locally storage and retrospective analysis.2. We enrolled 80 patients during the research period. Thirty eight episodes of infection occurred in 37 patients, including 29 episodes of late onset neonatal sepsis (8 episodes of lab-confirmed sepsis and 21 episodes of clinical sepsis),8 episodes of urine tract infection and 1 episodes of meningitis. The initial analysis result shows the patients with severe infection have decreased HRV and normal RRV.3. Forty eight patients’physiological data is relatively intact which can be used in the time-series analysis and met the standard of data mining. Six patients with sepsis (3 with lab confirmed sepsis and 3 with clinical sepsis) and 37 patients without sepsis were included. Average minutes of normal HRV per hour was compared between sepsis and non-sepsis patients. Relative alignment for sepsis patients considering average of 48 hours of data prior to diagnosis (30.42±6.06min/hour VS 33.6±6.02min/hour, P=0.18), average of 96 hours of data prior to diagnosis (30.2±6.02min/hour VS 33.6±6.02min/hour, P=0.17), average of all data prior to diagnosis (29.5±4.72min/hour VS 33.6±6.02min/hour, P=0.08) was compared with all other patients include all data. The average minutes of normal HRV per hour was lower with in patients with sepsis, but the difference have no statistic significance.4. Examine sepsis patients only with matched pairs analysis for sepsis patients (before and after sepsis treatment):1) when the发’Before’ period is average HRV (29.24 min) from all data leading up to point of sepsis diagnosis, ’After’ period is average HRV(33.62 min) from all data from 48 hours after day of sepsis diagnosis, there is a statistically significant increase in normal HRV after sepsis diagnosis as compared to before sepsis diagnosis (p=0.003), the mean difference is 4.38 additional minutes of normal HRV per hour after sepsis treatment(95%CI [2.1,6.7]).2) when the ’Before’ period is average HRV from 72 hours leading up to point of sepsis diagnosis, ’After’ period is average HRV from 72 hours after sepsis diagnosis, there is a statistically significant increase in normal HRV after sepsis diagnosis as compared to before sepsis diagnosis (p=0.04), the mean difference is 3.83 additional minutes of normal HRV(95%CI [1.4,5.9]).3) when ’Before’ period is average HRV from 72 hours leading up to point of sepsis diagnosis, ’After’ period is average HRV from 1 week after sepsis diagnosis, there is a statistically significant increase in normal HRV after sepsis diagnosis as compared to before sepsis diagnosis (p=0.006), the mean difference is 3.52 additional minutes of normal HRV(95%CI [1.3,5.8]).Conclusion1. The physiological data collection and storage platform (Artemis platform) was established in Chinese NICU base on the available bedside monitor and NICU environment.2. The stability and performance of the Artemis platform established in the NICU still need to be optimized and improved.3. HRV decreased in the sepsis patients according to our pilot study result.4. There may have a special time window to detect the decrease of HRV in the early stage of sepsis. Large cohort study is needed to further confirm the hypothesis.更多还原