【作者】 刘畅；

【导师】 李建国；

【作者基本信息】 武汉大学 ， 外科学， 2014， 博士

【摘要】 目的：呼吸机相关性肺炎(ventilator-associated pneumonia, VAP)的早期准确诊断非常困难。既往研究提示：对临床疑诊VAP患者的支气管肺泡灌洗液(bronchoalveolar lavage fluid, BALF)进行涂片、瑞氏-吉姆萨染色和显微镜镜检,在BALF中检测出含有胞内病原微生物(intracellular organisms, ICOs)的多型核粒细胞(polymorphonuclear leukocytes, PMNs)可能可以辅助VAP的早期诊断。但是该方法的有效性和最佳的诊断阈值仍未明确。本研究的目的是进一步评估该方法的诊断准确性,并对最佳诊断阈值进行探讨。方法：本研究为多中心前瞻临床观察研究,研究共纳入181名首次临床诊断为VAP的ICU患者,纳入研究的每名患者均接受纤维支气管镜检查以留取BALF样本。每名患者的BALF样本分别进行细菌定量培养、瑞氏-吉姆萨染色和革兰氏染色以明确细菌定量培养结果,含ICOs的PMNs是否存在及其比例(percentage of PMNs containing ICOs, PIC)和细菌形态学分类。纳入研究的患者按照研究设置的诊断标准明确最终诊断。分别以细菌定量培养、瑞氏-吉姆萨染色和革兰氏染色结果对VAP进行诊断并计算相应的诊断准确性。采用受试者操作特征曲线(receiver-operating characteristic curve, ROC)明确最佳的PIC诊断阈值。并对同一BALF样本,经细菌定量培养和革兰氏染色鉴定的细菌学分类结果进行比较。结果：最终162名患者被明确诊断,其中102名被诊断为VAP(VAP组),60名被诊断为非VAP(非VAP组)。ICOs在96.08%(98/102)的VAP患者和20.00%(12/60)的非VAP患者中被检出,VAP组的PIC显著高于非VAP组(9.53+6.65%比0.52±1.33%,P<0.01)。本研究中采用PIC诊断VAP的最佳诊断阈值是1.5%,以PIC>I.5%为标准诊断VAP的敏感性、特异性、阳性预测值和阴性预测值分别为94.12%、88.33%、93.20%和89.83%,ROC曲线下面积为0.956(95%置信区间,0.925-0.986,P<0.01)。以细菌定量培养阳性为标准诊断VAP的敏感性、特异性、阳性预测值和阴性预测值分别为65.69%,95.00%,95.71%和61.96%。以革兰氏染色检出细菌为标准诊断VAP的敏感性、特异性、阳性预测值和阴性预测值分别为70.59%,76.67%,83.72%和60.53%。同一BALF样本,经细菌定量培养和革兰氏染色分别鉴定的细菌学分类结果之间符合率较差,仅32.10%(52/162)完全符合,17.28%(28/162)部分符合。结论：对疑诊VAP患者,取BALF进行涂片、瑞氏-吉姆萨染色后,通过显微镜镜检ICOs可以快速辅助诊断VAP,以PIC>1.5%为诊断VAP的阈值具有良好的诊断准确性。由于革兰氏染色所鉴定的细菌学分类与最终细菌定量培养结果之间的符合率较差,革兰氏染色结果可能不足以指导VAP抗感染治疗时早期抗生素的选择。更多还原

【Abstract】 Background The presence of intracellular organisms (ICOs) in polymorphonuclear leukocytes obtained from bronchoalveolar lavage fluid (BALF) is a possible method for rapid diagnosis of ventilator-associated pneumonia (VAP). However, the validity of this diagnostic method remains controversial and the diagnostic thresholds reported by investigators were different. Our objective was to evaluate the accuracy of quantification of ICOs in BALF for the diagnosis of VAP, and to detect the best cutoff percentage of PMNs containing ICOs (PIC) in the microscopic examination of BALF for the diagnosis of VAP.Methods This was a prospective multi-center study conducted in4ICUs in Wuhan, China, which involving181patients suspected of first episode of VAP. BALF was obtained from all enrolled patients. The BALF samples underwent quantitative culture, cytological and bacteriological analysis to detect the culture results, PIC values and the morphological features of microorganisms. Definite diagnosis of VAP was based on pre-set criteria. The receiver-operating characteristic curve was used to detect the best cutoff point for PIC to diagnose VAP, and the diagnostic accuracy was calculated. Moreover, quantitative culture and Gram’s stain of BALF were adopted to diagnose VAP, and their diagnostic accuracy was evaluated as well.Results There were102patients definitely diagnosed with VAP (VAP group), and60patients definitely diagnosed without VAP (no VAP group). We found that ICOs were present in96.08%(98out of102) of VAP patients and20.00%(12out of60) of no VAP patients. The PICs were significantly higher (9.53±6.65%vs0.52±1.33%, P<0.01) in VAP group. In our study, the best cutoff point for PIC to diagnose VAP was1.5%,which had a sensitivity of94.12%,a specificity of88.33%, a positive predictive value (PPV) of93.20%and a negative predictive value (NPV) of89.83%.The area under the receiver-operating characteristic curve was0.956(95%confidence interval,0.925-0.986; P<0.01). When the positive quantitative culture results of BALF were used to diagnose VAP, the sensitivity, specificity, PPV and NPV were65.69%,95.00%,95.71%and61.96%, respectively. Whereas they were70.59%,76.67%,83.72%and60.53%, respectively, when the positive Gram’s stain results of BALF were used to diagnose VAP. The concordance between the results of Gram’s stain and quantitative cultures was poor, only32.10%(52out of162) was totally right, and17.28%(28out of162) was partially right.Conclusions PIC>1.5%has good diagnostic performance in the microscopic examination of BALF for the diagnosis of VAP. However, Gram’s stain is not reliable for the early adaptation of antibiotic therapy, due to the poor bacteriological predictive value.更多还原