Validity of emergency department syndromic surveillance for Influenza in children

描述(由调查者提供)：该研究的目标是描述来自急诊科(ED)的流感样疾病(ILI)数据在华盛顿州金县流行期预测实验室确认的儿童人群流感的有效性。具体目的是：1.评估和比较各种流感样疾病(ILI)症状检测实验室确诊流感的准确性；b)评估呼吸系统综合症检测实验室确诊流感的准确性。比较呼吸和ILI的准确性；以及c)(探索性分析)开发实验室确认的流感的症状预测模型，该模型将被转换为ILI分类器。目的2.描述ILI就诊与实验室监测阳性流感分离株之间的时间趋势，并进行比较。将以实验室监测为金标准，描述各种ILI综合征分类对生态准确性的影响。我们将随机选择在2000年至2005年期间到西雅图儿童医院急诊科就诊的18岁儿童的回顾性队列，以满足目标1的要求。主要主诉和诊断字段将用于将ILI和呼吸综合征的记录归类为阳性或阴性。测试的ILI和呼吸定义将代表目前在美国各地运行的定义。首席调查员还将测试通过探索性分析获得的新定义。鉴于存在ILI/呼吸综合征、行政和人口危险因素以及其他呼吸道感染，将使用Logistic回归对实验室确认的流感的概率进行建模。探索性分析中将使用逻辑回归，这是一种基于二进制编码风险因素的布尔组合来识别人口子集的方法。利用模拟的前瞻性分析，目标1的最佳ILI定义将适用于2005-2008年间出现的所有ED访问。所有就诊将被编码为ILI综合征阳性或阴性。将汇总ILI阳性访问，并使用动态线性模型(一种时间序列分析形式)估计频率，并与华盛顿大学临床病毒学实验室的阳性流感分离株频率进行比较。当观测值在某个阈值超过预期值时，将根据ED数据定义像差或信号。ED的信号将与实验室数据进行比较。该项目与日益重视改进流感监测系统以及儿童在其传播中所发挥的独特作用相一致。这项研究首次验证了针对单一实验室确认疾病的症状数据来源中的症状定义。

DESCRIPTION (provided by the investigator): The study objective is to describe the validity of influenza-like-illness (ILI) from emergency department (ED) data to predict laboratory-confirmed influenza in a pediatric population during epidemic periods in King County, Washington. The specific aims are to: Aim 1. a) Estimate and compare the accuracy of various influenza-like illness (ILI) syndromes for detecting laboratory-confirmed influenza; b) Estimate the accuracy of Respiratory Syndrome for detecting laboratory-confirmed influenza. Compare the accuracy of Respiratory with ILI; and c) (Exploratory Analysis) Develop a symptom prediction model for lab-confirmed influenza, which will be translated into an ILI classifier. Aim 2. Describe and compare temporal trends of ILI visits with positive influenza isolates from laboratory surveillance. The impact of various ILI syndrome classifications on ecological accuracy, using laboratory surveillance as a gold-standard, will be described. A retrospective cohort of children < 18 years of age who presented to the Seattle Children's Hospital ED during known influenza periods from 2000-2005 will be randomly selected to address Aim 1. Chief complaint and diagnoses fields will be used to categorize records as positive or negative for ILI and Respiratory Syndrome. ILI and Respiratory definitions tested will represent those currently used in operation across the U.S. The Principal Investigator will also test new definitions informed by exploratory analysis. Logistic regression will be used to model the probability of laboratory-confirmed influenza given presence of ILI/Respiratory Syndrome, administrative and demographic risk factors, and other respiratory infection. Logic regression, a method for identifying population subsets based on a Boolean combination of binary coded risk factors, will be used in the exploratory analysis. Using a simulated prospective analysis, optimal ILI definitions from Aim 1 will be applied to all ED visits presenting between 2005-2008. All visits will be coded as positive or negative for ILI syndrome. ILI positive visits will be aggregated, and frequencies estimated and compared with the frequency of positive influenza isolates from the UW Clinical Virology Laboratory using dynamic linear models, a form of time-series analysis. Aberrations, or signals, will be defined from ED data when observed values exceed expected values at some threshold. Signals from ED and laboratory data will be compared. The project is consistent with the increasing emphasis on improving surveillance systems for influenza and the unique role children play in its transmission. This study provides the first validation of a syndrome definition from syndromic data sources against a single, laboratory-confirmed disease.