中国不同患者人腺病毒检出率Meta分析

Meta analysis of detection rate of human adenovirus among different patients in China

目的了解人腺病毒(human adenovirus,HAdV)在不同疾病病原谱中的检出情况。方法对符合纳入标准的284个研究进行HAdV检出率的Meta分析,分析内容包括异质性检验、随机效应模型合并效应量、Meta回归、敏感性分析和发表偏倚。结果284个研究检出率波动较大、异质性较高,分别采用不同的方法降低或分析异质性:随机效应模型合并总检出率为4.21%(95%CI:3.80%~4.66%);亚组分析各组检出率分别为胃肠道症状疾病5.06%(95%CI:4.22%~6.06%)、呼吸道症状疾病3.97%(95%CI:3.53%~4.46%)、儿童4.50%(95%CI:4.06%~4.99%)、成人2.09%(95%CI:1.22%~3.56%)、抗原检测1.94%(95%CI:1.64%~2.30%)、核酸检测5.82%(95%CI:5.21%~6.49%)、抗体检测5.60%(95%CI:4.17%~7.48%)、中国东部地区3.67%(95%CI:3.18%~4.24%)、中部地区6.07%(95%CI:5.20%~7.08%)、西部地区5.00%(95%CI:4.03%~6.18%)、≤1000样本量6.10%(95%CI:5.39%~6.91%)、>1000样本量2.66%(95%CI:2.32%~3.05%);Meta回归分析有显著性的因素为疾病种类、年龄、检测方法、研究地区、样本量,调整R2=37.15%。结论HAdV检出率的研究,在异质性较大时,Meta分析对率的合并仍有待完善。不同疾病种类、不同年龄、检测方法、研究区域和样本量大小是引起异质性的部分原因。

Objective Human adenovirus(HAdV)can cause a variety of diseases,and sometimes leads to outbreaks of respiratory diseases,this study determines to understand the detection status of human adenovirus in the pathogenic spectrum of different diseases.Methods This study conducted Meta analysis on the detection rate of human adenovirus in 284 studies that met the inclusion criteria and included heterogeneity test,pooling rate by random effect model,Meta regression,sensitivity analysis and publication bias analysis.Results The detection rate varied greatly and heterogeneity of 284 studies was relatively large.Different method were used to reduce or analyze the heterogeneity:the total detection rate by random effect model was 4.21%(95%CI:3.80%~4.66%);the detection rates of each subgroup were gastrointestinal diseases 5.06%(95%CI:4.22%~6.06%),respiratory diseases 3.97%(95%CI:3.53%~4.46%),children 4.50%(95%CI:4.06%~4.99%),adults 2.09%(95%CI:1.22%~3.56%),antigen test 1.94%(95%CI:1.64%~2.30%),nucleic acid test 5.82%(95%CI:5.21%~6.49%),serum test 5.60%(95%CI:4.17%~7.48%),eastern 3.67%(95%CI:3.18%~4.24%),central 6.07%(95%CI:5.20%~7.08%),western 5.00%(95%CI:4.03%~6.18%),≤1000 sample size 6.10%(95%CI:5.39%~6.91%),>1000 sample size 2.66%(95%CI:2.32%~3.05%);the significant factors of Meta regression analysis were disease type,age,detection method,research area,sample size and adjusted R2=37.15%.Conclusions This study suggests that the Meta analysis needs to be improved to conduct pooling detection rate like HAdV with great heterogeneity.Different disease types,age,detection method,research area,sample size account for part of the resources for heterogeneity.