Ou Chun-quan, Deng Zhuo-hui, Yang Lin, Chen Ping-yan
Department of Biostatistics, School of Public Health and Tropical Medicine, Southern Medical University, Guangzhou, China.
Nan Fang Yi Ke Da Xue Xue Bao. 2008 Aug;28(8):1446-8.
To estimate the effect of influenza-like illness (ILI) on outpatient visits and assess its impact on public health.
We analyzed the data of weekly number of ILI and outpatient visits in Departments of Internal Medicine, Pediatrics and Emergency at two influenza surveillance hospitals during a period of 137 weeks in Guangzhou. Spectral analysis and time-series analysis were performed to evaluate the variation of outpatient visits over time. The predictive model was fitted with weekly outpatient visits as the dependent variable and weekly number of ILI as the independent variable. The optimal model was established according to the coefficient of determination, Akaike-information criterion and residual analysis. The validity of the model was assessed prospectively using the 31-week data that were not used for the model establishment.
The outpatient visits increased significantly over time and showed significant seasonality (P<0.001). A significant correlation was found between the weekly number of ILI and outpatient visits (r=0.568, P<0.001). The residuals of the fitted autoregression model were white-noise series and the coefficient of determination was 75% for the data used to establish the model and 56% for the subsequent 31-week data.
The autoregression model can be used to estimate the effect of weekly number of outpatient visits based on the weekly number of ILI and thus assess the effects of influenza on public health.
评估流感样疾病（ILI）对门诊就诊的影响，并评估其对公共卫生的影响。
我们分析了广州两家流感监测医院内科、儿科和急诊科在137周内每周ILI病例数和门诊就诊数据。采用频谱分析和时间序列分析来评估门诊就诊随时间的变化。以每周门诊就诊次数为因变量、每周ILI病例数为自变量建立预测模型。根据决定系数、赤池信息准则和残差分析建立最优模型。使用未用于模型建立的31周数据对模型的有效性进行前瞻性评估。
门诊就诊次数随时间显著增加且呈现明显的季节性（P<0.001）。每周ILI病例数与门诊就诊次数之间存在显著相关性（r=0.568，P<0.001）。拟合的自回归模型残差为白噪声序列，用于建立模型的数据的决定系数为75%，后续31周数据的决定系数为56%。
自回归模型可用于根据每周ILI病例数估计每周门诊就诊次数的影响，从而评估流感对公共卫生的影响。
